Terminal Server Performance

  • Thread starter Thread starter Grecko
  • Start date Start date
G

Grecko

Guest
We are having poor performance on our terminal server. We only have about
100 users on one server with 2 processors and 8 gigs of ram. The poor
performance seems to coincide with high/read writes on our C: drive. Could
the drive itself be the bottleneck or the network itself?...Thanks...Grecko
 
Re: Terminal Server Performance

Slowest part of any system is the hard drive.

what OS are you using? You say only 100 but I know companies that would
kill to get that many users on the system. What stress testing did you do
to the system originally to find out how many users it could handle? Did
you ever baseline it in any way? If you back the users down to say 85-90
does performance stay good?

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:031329B8-4AFC-4B87-A91A-66759D3694E4@microsoft.com...
> We are having poor performance on our terminal server. We only have about
> 100 users on one server with 2 processors and 8 gigs of ram. The poor
> performance seems to coincide with high/read writes on our C: drive.
> Could
> the drive itself be the bottleneck or the network
> itself?...Thanks...Grecko
 
Re: Terminal Server Performance

It's impossible to say here why the server is having performance
problems (but 100 concurrent sessions is quite a lot, depemding of
course on which applications the users are running).

You'll have to use Performance monitor to find out what the
bottleneck is. Since you have noticed high RW activity on the drive,
check the size and location of your swap file.
You can find some tools and guidelines here:
http://ts.veranoest.net/ts_performance.htm
_________________________________________________________
Vera Noest
MCSE, CCEA, Microsoft MVP - Terminal Server
TS troubleshooting: http://ts.veranoest.net
___ please respond in newsgroup, NOT by private email ___

=?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
03 jul 2008 in microsoft.public.windows.terminal_services:

> We are having poor performance on our terminal server. We only
> have about 100 users on one server with 2 processors and 8 gigs
> of ram. The poor performance seems to coincide with high/read
> writes on our C: drive. Could the drive itself be the
> bottleneck or the network itself?...Thanks...Grecko
 
Re: Terminal Server Performance

We are using Windows 2003 Enterprise. I am new to the company and as far as
I know they did not do any baseline testing.

It seems that the poor performance is happening when the read/writes are
high on the C: disk. Are the high read/writes caused by the network making
it take longer to accomplish them or does the high read/writes of the drive
happen separate from the network speed?

I guess what I am trying to figure out is if the drive issue could be
related to a network issue or is entirely separate. The network here is
setup incorrectly. They have static ips on machines, no quality of service
set, the network is not segmented, etc....Gregg


"Jeff Pitsch" wrote:

> Slowest part of any system is the hard drive.
>
> what OS are you using? You say only 100 but I know companies that would
> kill to get that many users on the system. What stress testing did you do
> to the system originally to find out how many users it could handle? Did
> you ever baseline it in any way? If you back the users down to say 85-90
> does performance stay good?
>
> Jeff Pitsch
> Microsoft MVP - Terminal Services
>
>
> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> news:031329B8-4AFC-4B87-A91A-66759D3694E4@microsoft.com...
> > We are having poor performance on our terminal server. We only have about
> > 100 users on one server with 2 processors and 8 gigs of ram. The poor
> > performance seems to coincide with high/read writes on our C: drive.
> > Could
> > the drive itself be the bottleneck or the network
> > itself?...Thanks...Grecko

>
>
>
 
Re: Terminal Server Performance

I checked the page file and its setup on drive C: for 400 to 1000 MB, on
drive F: for 12286 - 13386, drive H is a system managed drive, and driveE has
no paging file. Should we make the amount lower on C:. I am assuming it wont
roll to F until it uses the all page file on C:? Of course if this is the
case, then we would have a bottle neck on the C: drive. What would be the
best way to distribute the calls to the drives?...Gregg

"Vera Noest [MVP]" wrote:

> It's impossible to say here why the server is having performance
> problems (but 100 concurrent sessions is quite a lot, depemding of
> course on which applications the users are running).
>
> You'll have to use Performance monitor to find out what the
> bottleneck is. Since you have noticed high RW activity on the drive,
> check the size and location of your swap file.
> You can find some tools and guidelines here:
> http://ts.veranoest.net/ts_performance.htm
> _________________________________________________________
> Vera Noest
> MCSE, CCEA, Microsoft MVP - Terminal Server
> TS troubleshooting: http://ts.veranoest.net
> ___ please respond in newsgroup, NOT by private email ___
>
> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> 03 jul 2008 in microsoft.public.windows.terminal_services:
>
> > We are having poor performance on our terminal server. We only
> > have about 100 users on one server with 2 processors and 8 gigs
> > of ram. The poor performance seems to coincide with high/read
> > writes on our C: drive. Could the drive itself be the
> > bottleneck or the network itself?...Thanks...Grecko

>
 
Re: Terminal Server Performance

Page files should also be min/max the same so windows is not
growing/shrinking dynamically. this can add alot of overhead. Adding more
pagefile space would just worsen the situation.

both Vera and I have mentioned this but have you seriously considered that
you may have simply maxed out the server? Have you tried virtual memory
products from companies such as ThinPrint, RTO Soft, Provision Networks (I'm
an employee)? Those would cut down on your paging drasticallyl if there are
savings to be had.

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
>I checked the page file and its setup on drive C: for 400 to 1000 MB, on
> drive F: for 12286 - 13386, drive H is a system managed drive, and driveE
> has
> no paging file. Should we make the amount lower on C:. I am assuming it
> wont
> roll to F until it uses the all page file on C:? Of course if this is the
> case, then we would have a bottle neck on the C: drive. What would be the
> best way to distribute the calls to the drives?...Gregg
>
> "Vera Noest [MVP]" wrote:
>
>> It's impossible to say here why the server is having performance
>> problems (but 100 concurrent sessions is quite a lot, depemding of
>> course on which applications the users are running).
>>
>> You'll have to use Performance monitor to find out what the
>> bottleneck is. Since you have noticed high RW activity on the drive,
>> check the size and location of your swap file.
>> You can find some tools and guidelines here:
>> http://ts.veranoest.net/ts_performance.htm
>> _________________________________________________________
>> Vera Noest
>> MCSE, CCEA, Microsoft MVP - Terminal Server
>> TS troubleshooting: http://ts.veranoest.net
>> ___ please respond in newsgroup, NOT by private email ___
>>
>> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
>> 03 jul 2008 in microsoft.public.windows.terminal_services:
>>
>> > We are having poor performance on our terminal server. We only
>> > have about 100 users on one server with 2 processors and 8 gigs
>> > of ram. The poor performance seems to coincide with high/read
>> > writes on our C: drive. Could the drive itself be the
>> > bottleneck or the network itself?...Thanks...Grecko

>>
 
Re: Terminal Server Performance

We havent tried anything yet. Yes, we are probably going to add another
server to offload some of the users. We didnt want to do this until we had
figured out whats causing the problems.....network, hard-drive, etc...Grecko

"Jeff Pitsch" wrote:

> Page files should also be min/max the same so windows is not
> growing/shrinking dynamically. this can add alot of overhead. Adding more
> pagefile space would just worsen the situation.
>
> both Vera and I have mentioned this but have you seriously considered that
> you may have simply maxed out the server? Have you tried virtual memory
> products from companies such as ThinPrint, RTO Soft, Provision Networks (I'm
> an employee)? Those would cut down on your paging drasticallyl if there are
> savings to be had.
>
> Jeff Pitsch
> Microsoft MVP - Terminal Services
>
>
> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
> >I checked the page file and its setup on drive C: for 400 to 1000 MB, on
> > drive F: for 12286 - 13386, drive H is a system managed drive, and driveE
> > has
> > no paging file. Should we make the amount lower on C:. I am assuming it
> > wont
> > roll to F until it uses the all page file on C:? Of course if this is the
> > case, then we would have a bottle neck on the C: drive. What would be the
> > best way to distribute the calls to the drives?...Gregg
> >
> > "Vera Noest [MVP]" wrote:
> >
> >> It's impossible to say here why the server is having performance
> >> problems (but 100 concurrent sessions is quite a lot, depemding of
> >> course on which applications the users are running).
> >>
> >> You'll have to use Performance monitor to find out what the
> >> bottleneck is. Since you have noticed high RW activity on the drive,
> >> check the size and location of your swap file.
> >> You can find some tools and guidelines here:
> >> http://ts.veranoest.net/ts_performance.htm
> >> _________________________________________________________
> >> Vera Noest
> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> >> TS troubleshooting: http://ts.veranoest.net
> >> ___ please respond in newsgroup, NOT by private email ___
> >>
> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
> >>
> >> > We are having poor performance on our terminal server. We only
> >> > have about 100 users on one server with 2 processors and 8 gigs
> >> > of ram. The poor performance seems to coincide with high/read
> >> > writes on our C: drive. Could the drive itself be the
> >> > bottleneck or the network itself?...Thanks...Grecko
> >>

>
>
>
 
Re: Terminal Server Performance

I'm not sure how network would affect read/writes to the HD of the server.
The only way it could, that I can think of off hand, is if the network card
was bad and causing an unusual amount of interrupts forcing the processor to
work harder and forcing more read/writes as the cpu is backed up more and
more. What does the queue show for requests waiting to the hard drives? Is
CPU being spiked at all? How much memory is available on the servers when
this is happening?

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> We havent tried anything yet. Yes, we are probably going to add another
> server to offload some of the users. We didnt want to do this until we
> had
> figured out whats causing the problems.....network, hard-drive,
> etc...Grecko
>
> "Jeff Pitsch" wrote:
>
>> Page files should also be min/max the same so windows is not
>> growing/shrinking dynamically. this can add alot of overhead. Adding
>> more
>> pagefile space would just worsen the situation.
>>
>> both Vera and I have mentioned this but have you seriously considered
>> that
>> you may have simply maxed out the server? Have you tried virtual memory
>> products from companies such as ThinPrint, RTO Soft, Provision Networks
>> (I'm
>> an employee)? Those would cut down on your paging drasticallyl if there
>> are
>> savings to be had.
>>
>> Jeff Pitsch
>> Microsoft MVP - Terminal Services
>>
>>
>> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
>> >I checked the page file and its setup on drive C: for 400 to 1000 MB,
>> >on
>> > drive F: for 12286 - 13386, drive H is a system managed drive, and
>> > driveE
>> > has
>> > no paging file. Should we make the amount lower on C:. I am assuming
>> > it
>> > wont
>> > roll to F until it uses the all page file on C:? Of course if this is
>> > the
>> > case, then we would have a bottle neck on the C: drive. What would be
>> > the
>> > best way to distribute the calls to the drives?...Gregg
>> >
>> > "Vera Noest [MVP]" wrote:
>> >
>> >> It's impossible to say here why the server is having performance
>> >> problems (but 100 concurrent sessions is quite a lot, depemding of
>> >> course on which applications the users are running).
>> >>
>> >> You'll have to use Performance monitor to find out what the
>> >> bottleneck is. Since you have noticed high RW activity on the drive,
>> >> check the size and location of your swap file.
>> >> You can find some tools and guidelines here:
>> >> http://ts.veranoest.net/ts_performance.htm
>> >> _________________________________________________________
>> >> Vera Noest
>> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> >> TS troubleshooting: http://ts.veranoest.net
>> >> ___ please respond in newsgroup, NOT by private email ___
>> >>
>> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
>> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
>> >>
>> >> > We are having poor performance on our terminal server. We only
>> >> > have about 100 users on one server with 2 processors and 8 gigs
>> >> > of ram. The poor performance seems to coincide with high/read
>> >> > writes on our C: drive. Could the drive itself be the
>> >> > bottleneck or the network itself?...Thanks...Grecko
>> >>

>>
>>
>>
 
Re: Terminal Server Performance

There is a lot of memory still available on the server when the performance
is slow and no spikes in the processor. The only thing that indicates a
problem is the high read/writes in the perfmon. Our application people our
telling us that the bottle neck is in the network and that the read/writes
shouldnt have an effect on performance. Of course, we are seeing the opposite
of this. I am new to the environment and just trying to help each side. It
seems that it may be a little of both network and hard-drive read/writes as
best as I can tell. Any ideas?...Grecko

"Jeff Pitsch" wrote:

> I'm not sure how network would affect read/writes to the HD of the server.
> The only way it could, that I can think of off hand, is if the network card
> was bad and causing an unusual amount of interrupts forcing the processor to
> work harder and forcing more read/writes as the cpu is backed up more and
> more. What does the queue show for requests waiting to the hard drives? Is
> CPU being spiked at all? How much memory is available on the servers when
> this is happening?
>
> Jeff Pitsch
> Microsoft MVP - Terminal Services
>
>
> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> > We havent tried anything yet. Yes, we are probably going to add another
> > server to offload some of the users. We didnt want to do this until we
> > had
> > figured out whats causing the problems.....network, hard-drive,
> > etc...Grecko
> >
> > "Jeff Pitsch" wrote:
> >
> >> Page files should also be min/max the same so windows is not
> >> growing/shrinking dynamically. this can add alot of overhead. Adding
> >> more
> >> pagefile space would just worsen the situation.
> >>
> >> both Vera and I have mentioned this but have you seriously considered
> >> that
> >> you may have simply maxed out the server? Have you tried virtual memory
> >> products from companies such as ThinPrint, RTO Soft, Provision Networks
> >> (I'm
> >> an employee)? Those would cut down on your paging drasticallyl if there
> >> are
> >> savings to be had.
> >>
> >> Jeff Pitsch
> >> Microsoft MVP - Terminal Services
> >>
> >>
> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
> >> >I checked the page file and its setup on drive C: for 400 to 1000 MB,
> >> >on
> >> > drive F: for 12286 - 13386, drive H is a system managed drive, and
> >> > driveE
> >> > has
> >> > no paging file. Should we make the amount lower on C:. I am assuming
> >> > it
> >> > wont
> >> > roll to F until it uses the all page file on C:? Of course if this is
> >> > the
> >> > case, then we would have a bottle neck on the C: drive. What would be
> >> > the
> >> > best way to distribute the calls to the drives?...Gregg
> >> >
> >> > "Vera Noest [MVP]" wrote:
> >> >
> >> >> It's impossible to say here why the server is having performance
> >> >> problems (but 100 concurrent sessions is quite a lot, depemding of
> >> >> course on which applications the users are running).
> >> >>
> >> >> You'll have to use Performance monitor to find out what the
> >> >> bottleneck is. Since you have noticed high RW activity on the drive,
> >> >> check the size and location of your swap file.
> >> >> You can find some tools and guidelines here:
> >> >> http://ts.veranoest.net/ts_performance.htm
> >> >> _________________________________________________________
> >> >> Vera Noest
> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> >> >> TS troubleshooting: http://ts.veranoest.net
> >> >> ___ please respond in newsgroup, NOT by private email ___
> >> >>
> >> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> >> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
> >> >>
> >> >> > We are having poor performance on our terminal server. We only
> >> >> > have about 100 users on one server with 2 processors and 8 gigs
> >> >> > of ram. The poor performance seems to coincide with high/read
> >> >> > writes on our C: drive. Could the drive itself be the
> >> >> > bottleneck or the network itself?...Thanks...Grecko
> >> >>
> >>
> >>
> >>

>
>
>
 
Re: Terminal Server Performance

100 users does equal a lot fo traffic. Have you sniffed or analyzed the
network traffic? is the link saturated? Have you looked at the queue for
the hard drives to see if commands are lining up during peak periods? What
RAID config are you using?

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> There is a lot of memory still available on the server when the
> performance
> is slow and no spikes in the processor. The only thing that indicates a
> problem is the high read/writes in the perfmon. Our application people
> our
> telling us that the bottle neck is in the network and that the read/writes
> shouldnt have an effect on performance. Of course, we are seeing the
> opposite
> of this. I am new to the environment and just trying to help each side.
> It
> seems that it may be a little of both network and hard-drive read/writes
> as
> best as I can tell. Any ideas?...Grecko
>
> "Jeff Pitsch" wrote:
>
>> I'm not sure how network would affect read/writes to the HD of the
>> server.
>> The only way it could, that I can think of off hand, is if the network
>> card
>> was bad and causing an unusual amount of interrupts forcing the processor
>> to
>> work harder and forcing more read/writes as the cpu is backed up more and
>> more. What does the queue show for requests waiting to the hard drives?
>> Is
>> CPU being spiked at all? How much memory is available on the servers
>> when
>> this is happening?
>>
>> Jeff Pitsch
>> Microsoft MVP - Terminal Services
>>
>>
>> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
>> > We havent tried anything yet. Yes, we are probably going to add
>> > another
>> > server to offload some of the users. We didnt want to do this until we
>> > had
>> > figured out whats causing the problems.....network, hard-drive,
>> > etc...Grecko
>> >
>> > "Jeff Pitsch" wrote:
>> >
>> >> Page files should also be min/max the same so windows is not
>> >> growing/shrinking dynamically. this can add alot of overhead. Adding
>> >> more
>> >> pagefile space would just worsen the situation.
>> >>
>> >> both Vera and I have mentioned this but have you seriously considered
>> >> that
>> >> you may have simply maxed out the server? Have you tried virtual
>> >> memory
>> >> products from companies such as ThinPrint, RTO Soft, Provision
>> >> Networks
>> >> (I'm
>> >> an employee)? Those would cut down on your paging drasticallyl if
>> >> there
>> >> are
>> >> savings to be had.
>> >>
>> >> Jeff Pitsch
>> >> Microsoft MVP - Terminal Services
>> >>
>> >>
>> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
>> >> >I checked the page file and its setup on drive C: for 400 to 1000
>> >> >MB,
>> >> >on
>> >> > drive F: for 12286 - 13386, drive H is a system managed drive, and
>> >> > driveE
>> >> > has
>> >> > no paging file. Should we make the amount lower on C:. I am
>> >> > assuming
>> >> > it
>> >> > wont
>> >> > roll to F until it uses the all page file on C:? Of course if this
>> >> > is
>> >> > the
>> >> > case, then we would have a bottle neck on the C: drive. What would
>> >> > be
>> >> > the
>> >> > best way to distribute the calls to the drives?...Gregg
>> >> >
>> >> > "Vera Noest [MVP]" wrote:
>> >> >
>> >> >> It's impossible to say here why the server is having performance
>> >> >> problems (but 100 concurrent sessions is quite a lot, depemding of
>> >> >> course on which applications the users are running).
>> >> >>
>> >> >> You'll have to use Performance monitor to find out what the
>> >> >> bottleneck is. Since you have noticed high RW activity on the
>> >> >> drive,
>> >> >> check the size and location of your swap file.
>> >> >> You can find some tools and guidelines here:
>> >> >> http://ts.veranoest.net/ts_performance.htm
>> >> >> _________________________________________________________
>> >> >> Vera Noest
>> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> >> >> TS troubleshooting: http://ts.veranoest.net
>> >> >> ___ please respond in newsgroup, NOT by private email ___
>> >> >>
>> >> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
>> >> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
>> >> >>
>> >> >> > We are having poor performance on our terminal server. We only
>> >> >> > have about 100 users on one server with 2 processors and 8 gigs
>> >> >> > of ram. The poor performance seems to coincide with high/read
>> >> >> > writes on our C: drive. Could the drive itself be the
>> >> >> > bottleneck or the network itself?...Thanks...Grecko
>> >> >>
>> >>
>> >>
>> >>

>>
>>
>>
 
Re: Terminal Server Performance

No we havent sniffed or analyzed the network traffic. Like I said I know
they have network issues because they still have static ips, there is no
segments in the network, etc. What tool would you recommend to analyze the
traffic? What do you mean by the link saturated? We havent looked at the
queue yet during peak period...just the reads/writes. What would the queue
indicate?...We are using RAID1 mirroring...Grecko

"Jeff Pitsch" wrote:

> 100 users does equal a lot fo traffic. Have you sniffed or analyzed the
> network traffic? is the link saturated? Have you looked at the queue for
> the hard drives to see if commands are lining up during peak periods? What
> RAID config are you using?
>
> Jeff Pitsch
> Microsoft MVP - Terminal Services
>
>
> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> > There is a lot of memory still available on the server when the
> > performance
> > is slow and no spikes in the processor. The only thing that indicates a
> > problem is the high read/writes in the perfmon. Our application people
> > our
> > telling us that the bottle neck is in the network and that the read/writes
> > shouldnt have an effect on performance. Of course, we are seeing the
> > opposite
> > of this. I am new to the environment and just trying to help each side.
> > It
> > seems that it may be a little of both network and hard-drive read/writes
> > as
> > best as I can tell. Any ideas?...Grecko
> >
> > "Jeff Pitsch" wrote:
> >
> >> I'm not sure how network would affect read/writes to the HD of the
> >> server.
> >> The only way it could, that I can think of off hand, is if the network
> >> card
> >> was bad and causing an unusual amount of interrupts forcing the processor
> >> to
> >> work harder and forcing more read/writes as the cpu is backed up more and
> >> more. What does the queue show for requests waiting to the hard drives?
> >> Is
> >> CPU being spiked at all? How much memory is available on the servers
> >> when
> >> this is happening?
> >>
> >> Jeff Pitsch
> >> Microsoft MVP - Terminal Services
> >>
> >>
> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> >> > We havent tried anything yet. Yes, we are probably going to add
> >> > another
> >> > server to offload some of the users. We didnt want to do this until we
> >> > had
> >> > figured out whats causing the problems.....network, hard-drive,
> >> > etc...Grecko
> >> >
> >> > "Jeff Pitsch" wrote:
> >> >
> >> >> Page files should also be min/max the same so windows is not
> >> >> growing/shrinking dynamically. this can add alot of overhead. Adding
> >> >> more
> >> >> pagefile space would just worsen the situation.
> >> >>
> >> >> both Vera and I have mentioned this but have you seriously considered
> >> >> that
> >> >> you may have simply maxed out the server? Have you tried virtual
> >> >> memory
> >> >> products from companies such as ThinPrint, RTO Soft, Provision
> >> >> Networks
> >> >> (I'm
> >> >> an employee)? Those would cut down on your paging drasticallyl if
> >> >> there
> >> >> are
> >> >> savings to be had.
> >> >>
> >> >> Jeff Pitsch
> >> >> Microsoft MVP - Terminal Services
> >> >>
> >> >>
> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
> >> >> >I checked the page file and its setup on drive C: for 400 to 1000
> >> >> >MB,
> >> >> >on
> >> >> > drive F: for 12286 - 13386, drive H is a system managed drive, and
> >> >> > driveE
> >> >> > has
> >> >> > no paging file. Should we make the amount lower on C:. I am
> >> >> > assuming
> >> >> > it
> >> >> > wont
> >> >> > roll to F until it uses the all page file on C:? Of course if this
> >> >> > is
> >> >> > the
> >> >> > case, then we would have a bottle neck on the C: drive. What would
> >> >> > be
> >> >> > the
> >> >> > best way to distribute the calls to the drives?...Gregg
> >> >> >
> >> >> > "Vera Noest [MVP]" wrote:
> >> >> >
> >> >> >> It's impossible to say here why the server is having performance
> >> >> >> problems (but 100 concurrent sessions is quite a lot, depemding of
> >> >> >> course on which applications the users are running).
> >> >> >>
> >> >> >> You'll have to use Performance monitor to find out what the
> >> >> >> bottleneck is. Since you have noticed high RW activity on the
> >> >> >> drive,
> >> >> >> check the size and location of your swap file.
> >> >> >> You can find some tools and guidelines here:
> >> >> >> http://ts.veranoest.net/ts_performance.htm
> >> >> >> _________________________________________________________
> >> >> >> Vera Noest
> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> >> >> >> TS troubleshooting: http://ts.veranoest.net
> >> >> >> ___ please respond in newsgroup, NOT by private email ___
> >> >> >>
> >> >> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> >> >> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
> >> >> >>
> >> >> >> > We are having poor performance on our terminal server. We only
> >> >> >> > have about 100 users on one server with 2 processors and 8 gigs
> >> >> >> > of ram. The poor performance seems to coincide with high/read
> >> >> >> > writes on our C: drive. Could the drive itself be the
> >> >> >> > bottleneck or the network itself?...Thanks...Grecko
> >> >> >>
> >> >>
> >> >>
> >> >>
> >>
> >>
> >>

>
>
>
 
Re: Terminal Server Performance

Read/Writes themselves aren't really indicitive of a problem unless the
queue is building up. If there is no queue then your disks probably aren't
a problem. In other words, no commands are waiting for the disk/channel to
free up.

As for the tool, there's a bunch out there but the free ones slip my mind.
almost midnight and I'm tired :( but you really are looking to see if the,
I'm assuming, 100meg connection is simply saturated tot he terminal server.
Out of curiousity are they using a hub or switch for all those single
segment connections?

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
> No we havent sniffed or analyzed the network traffic. Like I said I know
> they have network issues because they still have static ips, there is no
> segments in the network, etc. What tool would you recommend to analyze the
> traffic? What do you mean by the link saturated? We havent looked at the
> queue yet during peak period...just the reads/writes. What would the
> queue
> indicate?...We are using RAID1 mirroring...Grecko
>
> "Jeff Pitsch" wrote:
>
>> 100 users does equal a lot fo traffic. Have you sniffed or analyzed the
>> network traffic? is the link saturated? Have you looked at the queue
>> for
>> the hard drives to see if commands are lining up during peak periods?
>> What
>> RAID config are you using?
>>
>> Jeff Pitsch
>> Microsoft MVP - Terminal Services
>>
>>
>> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
>> > There is a lot of memory still available on the server when the
>> > performance
>> > is slow and no spikes in the processor. The only thing that indicates
>> > a
>> > problem is the high read/writes in the perfmon. Our application people
>> > our
>> > telling us that the bottle neck is in the network and that the
>> > read/writes
>> > shouldnt have an effect on performance. Of course, we are seeing the
>> > opposite
>> > of this. I am new to the environment and just trying to help each
>> > side.
>> > It
>> > seems that it may be a little of both network and hard-drive
>> > read/writes
>> > as
>> > best as I can tell. Any ideas?...Grecko
>> >
>> > "Jeff Pitsch" wrote:
>> >
>> >> I'm not sure how network would affect read/writes to the HD of the
>> >> server.
>> >> The only way it could, that I can think of off hand, is if the network
>> >> card
>> >> was bad and causing an unusual amount of interrupts forcing the
>> >> processor
>> >> to
>> >> work harder and forcing more read/writes as the cpu is backed up more
>> >> and
>> >> more. What does the queue show for requests waiting to the hard
>> >> drives?
>> >> Is
>> >> CPU being spiked at all? How much memory is available on the servers
>> >> when
>> >> this is happening?
>> >>
>> >> Jeff Pitsch
>> >> Microsoft MVP - Terminal Services
>> >>
>> >>
>> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
>> >> > We havent tried anything yet. Yes, we are probably going to add
>> >> > another
>> >> > server to offload some of the users. We didnt want to do this until
>> >> > we
>> >> > had
>> >> > figured out whats causing the problems.....network, hard-drive,
>> >> > etc...Grecko
>> >> >
>> >> > "Jeff Pitsch" wrote:
>> >> >
>> >> >> Page files should also be min/max the same so windows is not
>> >> >> growing/shrinking dynamically. this can add alot of overhead.
>> >> >> Adding
>> >> >> more
>> >> >> pagefile space would just worsen the situation.
>> >> >>
>> >> >> both Vera and I have mentioned this but have you seriously
>> >> >> considered
>> >> >> that
>> >> >> you may have simply maxed out the server? Have you tried virtual
>> >> >> memory
>> >> >> products from companies such as ThinPrint, RTO Soft, Provision
>> >> >> Networks
>> >> >> (I'm
>> >> >> an employee)? Those would cut down on your paging drasticallyl if
>> >> >> there
>> >> >> are
>> >> >> savings to be had.
>> >> >>
>> >> >> Jeff Pitsch
>> >> >> Microsoft MVP - Terminal Services
>> >> >>
>> >> >>
>> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
>> >> >> >I checked the page file and its setup on drive C: for 400 to 1000
>> >> >> >MB,
>> >> >> >on
>> >> >> > drive F: for 12286 - 13386, drive H is a system managed drive,
>> >> >> > and
>> >> >> > driveE
>> >> >> > has
>> >> >> > no paging file. Should we make the amount lower on C:. I am
>> >> >> > assuming
>> >> >> > it
>> >> >> > wont
>> >> >> > roll to F until it uses the all page file on C:? Of course if
>> >> >> > this
>> >> >> > is
>> >> >> > the
>> >> >> > case, then we would have a bottle neck on the C: drive. What
>> >> >> > would
>> >> >> > be
>> >> >> > the
>> >> >> > best way to distribute the calls to the drives?...Gregg
>> >> >> >
>> >> >> > "Vera Noest [MVP]" wrote:
>> >> >> >
>> >> >> >> It's impossible to say here why the server is having performance
>> >> >> >> problems (but 100 concurrent sessions is quite a lot, depemding
>> >> >> >> of
>> >> >> >> course on which applications the users are running).
>> >> >> >>
>> >> >> >> You'll have to use Performance monitor to find out what the
>> >> >> >> bottleneck is. Since you have noticed high RW activity on the
>> >> >> >> drive,
>> >> >> >> check the size and location of your swap file.
>> >> >> >> You can find some tools and guidelines here:
>> >> >> >> http://ts.veranoest.net/ts_performance.htm
>> >> >> >> _________________________________________________________
>> >> >> >> Vera Noest
>> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> >> >> >> TS troubleshooting: http://ts.veranoest.net
>> >> >> >> ___ please respond in newsgroup, NOT by private email ___
>> >> >> >>
>> >> >> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
>> >> >> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
>> >> >> >>
>> >> >> >> > We are having poor performance on our terminal server. We
>> >> >> >> > only
>> >> >> >> > have about 100 users on one server with 2 processors and 8
>> >> >> >> > gigs
>> >> >> >> > of ram. The poor performance seems to coincide with high/read
>> >> >> >> > writes on our C: drive. Could the drive itself be the
>> >> >> >> > bottleneck or the network itself?...Thanks...Grecko
>> >> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >>
>> >>
>> >>

>>
>>
>>
 
Re: Terminal Server Performance

I hope you had a good 4th. I will have to look at the disk queue on Monday.

We are using switches and the server has a 1 gbps card in it connecting it
to the switch. I already told them they need to get a fiber connection from
the server to the switch. Is there anyway to tell from the Windows Task
Manager if the network connection is saturated? What is an acceptable range
for the percentage in the network utilization?... I really appreciate the
help...Grecko

"Jeff Pitsch" wrote:

> Read/Writes themselves aren't really indicitive of a problem unless the
> queue is building up. If there is no queue then your disks probably aren't
> a problem. In other words, no commands are waiting for the disk/channel to
> free up.
>
> As for the tool, there's a bunch out there but the free ones slip my mind.
> almost midnight and I'm tired :( but you really are looking to see if the,
> I'm assuming, 100meg connection is simply saturated tot he terminal server.
> Out of curiousity are they using a hub or switch for all those single
> segment connections?
>
> Jeff Pitsch
> Microsoft MVP - Terminal Services
>
>
> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
> > No we havent sniffed or analyzed the network traffic. Like I said I know
> > they have network issues because they still have static ips, there is no
> > segments in the network, etc. What tool would you recommend to analyze the
> > traffic? What do you mean by the link saturated? We havent looked at the
> > queue yet during peak period...just the reads/writes. What would the
> > queue
> > indicate?...We are using RAID1 mirroring...Grecko
> >
> > "Jeff Pitsch" wrote:
> >
> >> 100 users does equal a lot fo traffic. Have you sniffed or analyzed the
> >> network traffic? is the link saturated? Have you looked at the queue
> >> for
> >> the hard drives to see if commands are lining up during peak periods?
> >> What
> >> RAID config are you using?
> >>
> >> Jeff Pitsch
> >> Microsoft MVP - Terminal Services
> >>
> >>
> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> >> > There is a lot of memory still available on the server when the
> >> > performance
> >> > is slow and no spikes in the processor. The only thing that indicates
> >> > a
> >> > problem is the high read/writes in the perfmon. Our application people
> >> > our
> >> > telling us that the bottle neck is in the network and that the
> >> > read/writes
> >> > shouldnt have an effect on performance. Of course, we are seeing the
> >> > opposite
> >> > of this. I am new to the environment and just trying to help each
> >> > side.
> >> > It
> >> > seems that it may be a little of both network and hard-drive
> >> > read/writes
> >> > as
> >> > best as I can tell. Any ideas?...Grecko
> >> >
> >> > "Jeff Pitsch" wrote:
> >> >
> >> >> I'm not sure how network would affect read/writes to the HD of the
> >> >> server.
> >> >> The only way it could, that I can think of off hand, is if the network
> >> >> card
> >> >> was bad and causing an unusual amount of interrupts forcing the
> >> >> processor
> >> >> to
> >> >> work harder and forcing more read/writes as the cpu is backed up more
> >> >> and
> >> >> more. What does the queue show for requests waiting to the hard
> >> >> drives?
> >> >> Is
> >> >> CPU being spiked at all? How much memory is available on the servers
> >> >> when
> >> >> this is happening?
> >> >>
> >> >> Jeff Pitsch
> >> >> Microsoft MVP - Terminal Services
> >> >>
> >> >>
> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> >> >> > We havent tried anything yet. Yes, we are probably going to add
> >> >> > another
> >> >> > server to offload some of the users. We didnt want to do this until
> >> >> > we
> >> >> > had
> >> >> > figured out whats causing the problems.....network, hard-drive,
> >> >> > etc...Grecko
> >> >> >
> >> >> > "Jeff Pitsch" wrote:
> >> >> >
> >> >> >> Page files should also be min/max the same so windows is not
> >> >> >> growing/shrinking dynamically. this can add alot of overhead.
> >> >> >> Adding
> >> >> >> more
> >> >> >> pagefile space would just worsen the situation.
> >> >> >>
> >> >> >> both Vera and I have mentioned this but have you seriously
> >> >> >> considered
> >> >> >> that
> >> >> >> you may have simply maxed out the server? Have you tried virtual
> >> >> >> memory
> >> >> >> products from companies such as ThinPrint, RTO Soft, Provision
> >> >> >> Networks
> >> >> >> (I'm
> >> >> >> an employee)? Those would cut down on your paging drasticallyl if
> >> >> >> there
> >> >> >> are
> >> >> >> savings to be had.
> >> >> >>
> >> >> >> Jeff Pitsch
> >> >> >> Microsoft MVP - Terminal Services
> >> >> >>
> >> >> >>
> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com...
> >> >> >> >I checked the page file and its setup on drive C: for 400 to 1000
> >> >> >> >MB,
> >> >> >> >on
> >> >> >> > drive F: for 12286 - 13386, drive H is a system managed drive,
> >> >> >> > and
> >> >> >> > driveE
> >> >> >> > has
> >> >> >> > no paging file. Should we make the amount lower on C:. I am
> >> >> >> > assuming
> >> >> >> > it
> >> >> >> > wont
> >> >> >> > roll to F until it uses the all page file on C:? Of course if
> >> >> >> > this
> >> >> >> > is
> >> >> >> > the
> >> >> >> > case, then we would have a bottle neck on the C: drive. What
> >> >> >> > would
> >> >> >> > be
> >> >> >> > the
> >> >> >> > best way to distribute the calls to the drives?...Gregg
> >> >> >> >
> >> >> >> > "Vera Noest [MVP]" wrote:
> >> >> >> >
> >> >> >> >> It's impossible to say here why the server is having performance
> >> >> >> >> problems (but 100 concurrent sessions is quite a lot, depemding
> >> >> >> >> of
> >> >> >> >> course on which applications the users are running).
> >> >> >> >>
> >> >> >> >> You'll have to use Performance monitor to find out what the
> >> >> >> >> bottleneck is. Since you have noticed high RW activity on the
> >> >> >> >> drive,
> >> >> >> >> check the size and location of your swap file.
> >> >> >> >> You can find some tools and guidelines here:
> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
> >> >> >> >> _________________________________________________________
> >> >> >> >> Vera Noest
> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
> >> >> >> >> ___ please respond in newsgroup, NOT by private email ___
> >> >> >> >>
> >> >> >> >> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> >> >> >> >> 03 jul 2008 in microsoft.public.windows.terminal_services:
> >> >> >> >>
> >> >> >> >> > We are having poor performance on our terminal server. We
> >> >> >> >> > only
> >> >> >> >> > have about 100 users on one server with 2 processors and 8
> >> >> >> >> > gigs
> >> >> >> >> > of ram. The poor performance seems to coincide with high/read
> >> >> >> >> > writes on our C: drive. Could the drive itself be the
> >> >> >> >> > bottleneck or the network itself?...Thanks...Grecko
> >> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >>
> >> >>
> >> >>
> >>
> >>
> >>

>
>
>
 
Re: Terminal Server Performance

No, not from Task Manager. You'll need Network Monitor. Check if
this helps:

How can I measure RDP bandwidth usage?
http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwidth

_________________________________________________________
Vera Noest
MCSE, CCEA, Microsoft MVP - Terminal Server
TS troubleshooting: http://ts.veranoest.net
___ please respond in newsgroup, NOT by private email ___

=?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
05 jul 2008 in microsoft.public.windows.terminal_services:

> I hope you had a good 4th. I will have to look at the disk
> queue on Monday.
>
> We are using switches and the server has a 1 gbps card in it
> connecting it to the switch. I already told them they need to
> get a fiber connection from the server to the switch. Is there
> anyway to tell from the Windows Task Manager if the network
> connection is saturated? What is an acceptable range for the
> percentage in the network utilization?... I really appreciate
> the help...Grecko
>
> "Jeff Pitsch" wrote:
>
>> Read/Writes themselves aren't really indicitive of a problem
>> unless the queue is building up. If there is no queue then
>> your disks probably aren't a problem. In other words, no
>> commands are waiting for the disk/channel to free up.
>>
>> As for the tool, there's a bunch out there but the free ones
>> slip my mind. almost midnight and I'm tired :( but you really
>> are looking to see if the, I'm assuming, 100meg connection is
>> simply saturated tot he terminal server. Out of curiousity are
>> they using a hub or switch for all those single segment
>> connections?
>>
>> Jeff Pitsch
>> Microsoft MVP - Terminal Services
>>
>>
>> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
>> > No we havent sniffed or analyzed the network traffic. Like I
>> > said I know they have network issues because they still have
>> > static ips, there is no segments in the network, etc. What
>> > tool would you recommend to analyze the traffic? What do you
>> > mean by the link saturated? We havent looked at the queue
>> > yet during peak period...just the reads/writes. What would
>> > the queue
>> > indicate?...We are using RAID1 mirroring...Grecko
>> >
>> > "Jeff Pitsch" wrote:
>> >
>> >> 100 users does equal a lot fo traffic. Have you sniffed or
>> >> analyzed the network traffic? is the link saturated? Have
>> >> you looked at the queue for
>> >> the hard drives to see if commands are lining up during peak
>> >> periods? What
>> >> RAID config are you using?
>> >>
>> >> Jeff Pitsch
>> >> Microsoft MVP - Terminal Services
>> >>
>> >>
>> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
>> >> > There is a lot of memory still available on the server
>> >> > when the performance
>> >> > is slow and no spikes in the processor. The only thing
>> >> > that indicates a
>> >> > problem is the high read/writes in the perfmon. Our
>> >> > application people our
>> >> > telling us that the bottle neck is in the network and that
>> >> > the read/writes
>> >> > shouldnt have an effect on performance. Of course, we are
>> >> > seeing the opposite
>> >> > of this. I am new to the environment and just trying to
>> >> > help each side.
>> >> > It
>> >> > seems that it may be a little of both network and
>> >> > hard-drive read/writes
>> >> > as
>> >> > best as I can tell. Any ideas?...Grecko
>> >> >
>> >> > "Jeff Pitsch" wrote:
>> >> >
>> >> >> I'm not sure how network would affect read/writes to the
>> >> >> HD of the server.
>> >> >> The only way it could, that I can think of off hand, is
>> >> >> if the network card
>> >> >> was bad and causing an unusual amount of interrupts
>> >> >> forcing the processor
>> >> >> to
>> >> >> work harder and forcing more read/writes as the cpu is
>> >> >> backed up more and
>> >> >> more. What does the queue show for requests waiting to
>> >> >> the hard drives?
>> >> >> Is
>> >> >> CPU being spiked at all? How much memory is available on
>> >> >> the servers when
>> >> >> this is happening?
>> >> >>
>> >> >> Jeff Pitsch
>> >> >> Microsoft MVP - Terminal Services
>> >> >>
>> >> >>
>> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> >> >> message
>> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
>> >> >> > We havent tried anything yet. Yes, we are probably
>> >> >> > going to add another
>> >> >> > server to offload some of the users. We didnt want to
>> >> >> > do this until we
>> >> >> > had
>> >> >> > figured out whats causing the problems.....network,
>> >> >> > hard-drive, etc...Grecko
>> >> >> >
>> >> >> > "Jeff Pitsch" wrote:
>> >> >> >
>> >> >> >> Page files should also be min/max the same so windows
>> >> >> >> is not growing/shrinking dynamically. this can add
>> >> >> >> alot of overhead. Adding
>> >> >> >> more
>> >> >> >> pagefile space would just worsen the situation.
>> >> >> >>
>> >> >> >> both Vera and I have mentioned this but have you
>> >> >> >> seriously considered
>> >> >> >> that
>> >> >> >> you may have simply maxed out the server? Have you
>> >> >> >> tried virtual memory
>> >> >> >> products from companies such as ThinPrint, RTO Soft,
>> >> >> >> Provision Networks
>> >> >> >> (I'm
>> >> >> >> an employee)? Those would cut down on your paging
>> >> >> >> drasticallyl if there
>> >> >> >> are
>> >> >> >> savings to be had.
>> >> >> >>
>> >> >> >> Jeff Pitsch
>> >> >> >> Microsoft MVP - Terminal Services
>> >> >> >>
>> >> >> >>
>> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> >> >> >> message
>> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com
>> >> >> >> ...
>> >> >> >> >I checked the page file and its setup on drive C: for
>> >> >> >> > 400 to 1000 MB,
>> >> >> >> >on
>> >> >> >> > drive F: for 12286 - 13386, drive H is a system
>> >> >> >> > managed drive, and
>> >> >> >> > driveE
>> >> >> >> > has
>> >> >> >> > no paging file. Should we make the amount lower on
>> >> >> >> > C:. I am assuming
>> >> >> >> > it
>> >> >> >> > wont
>> >> >> >> > roll to F until it uses the all page file on C:? Of
>> >> >> >> > course if this
>> >> >> >> > is
>> >> >> >> > the
>> >> >> >> > case, then we would have a bottle neck on the C:
>> >> >> >> > drive. What would
>> >> >> >> > be
>> >> >> >> > the
>> >> >> >> > best way to distribute the calls to the
>> >> >> >> > drives?...Gregg
>> >> >> >> >
>> >> >> >> > "Vera Noest [MVP]" wrote:
>> >> >> >> >
>> >> >> >> >> It's impossible to say here why the server is
>> >> >> >> >> having performance problems (but 100 concurrent
>> >> >> >> >> sessions is quite a lot, depemding of
>> >> >> >> >> course on which applications the users are
>> >> >> >> >> running).
>> >> >> >> >>
>> >> >> >> >> You'll have to use Performance monitor to find out
>> >> >> >> >> what the bottleneck is. Since you have noticed high
>> >> >> >> >> RW activity on the drive,
>> >> >> >> >> check the size and location of your swap file.
>> >> >> >> >> You can find some tools and guidelines here:
>> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
>> >> >> >> >> ____________________________________________________
>> >> >> >> >> _____ Vera Noest
>> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
>> >> >> >> >> ___ please respond in newsgroup, NOT by private
>> >> >> >> >> email ___
>> >> >> >> >>
>> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
>> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
>> >> >> >> >> 03 jul 2008 in
>> >> >> >> >> microsoft.public.windows.terminal_services:
>> >> >> >> >>
>> >> >> >> >> > We are having poor performance on our terminal
>> >> >> >> >> > server. We only
>> >> >> >> >> > have about 100 users on one server with 2
>> >> >> >> >> > processors and 8 gigs
>> >> >> >> >> > of ram. The poor performance seems to coincide
>> >> >> >> >> > with high/read writes on our C: drive. Could the
>> >> >> >> >> > drive itself be the bottleneck or the network
>> >> >> >> >> > itself?...Thanks...Grecko
 
Re: Terminal Server Performance

Thanks. I will try that out. So far today, the system is running smoothly.
I will have to analyze the data when the server starts going slow again. Any
other free programs you know of to monitor the traffic?...Grecko

"Vera Noest [MVP]" wrote:

> No, not from Task Manager. You'll need Network Monitor. Check if
> this helps:
>
> How can I measure RDP bandwidth usage?
> http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwidth
>
> _________________________________________________________
> Vera Noest
> MCSE, CCEA, Microsoft MVP - Terminal Server
> TS troubleshooting: http://ts.veranoest.net
> ___ please respond in newsgroup, NOT by private email ___
>
> =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> 05 jul 2008 in microsoft.public.windows.terminal_services:
>
> > I hope you had a good 4th. I will have to look at the disk
> > queue on Monday.
> >
> > We are using switches and the server has a 1 gbps card in it
> > connecting it to the switch. I already told them they need to
> > get a fiber connection from the server to the switch. Is there
> > anyway to tell from the Windows Task Manager if the network
> > connection is saturated? What is an acceptable range for the
> > percentage in the network utilization?... I really appreciate
> > the help...Grecko
> >
> > "Jeff Pitsch" wrote:
> >
> >> Read/Writes themselves aren't really indicitive of a problem
> >> unless the queue is building up. If there is no queue then
> >> your disks probably aren't a problem. In other words, no
> >> commands are waiting for the disk/channel to free up.
> >>
> >> As for the tool, there's a bunch out there but the free ones
> >> slip my mind. almost midnight and I'm tired :( but you really
> >> are looking to see if the, I'm assuming, 100meg connection is
> >> simply saturated tot he terminal server. Out of curiousity are
> >> they using a hub or switch for all those single segment
> >> connections?
> >>
> >> Jeff Pitsch
> >> Microsoft MVP - Terminal Services
> >>
> >>
> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
> >> > No we havent sniffed or analyzed the network traffic. Like I
> >> > said I know they have network issues because they still have
> >> > static ips, there is no segments in the network, etc. What
> >> > tool would you recommend to analyze the traffic? What do you
> >> > mean by the link saturated? We havent looked at the queue
> >> > yet during peak period...just the reads/writes. What would
> >> > the queue
> >> > indicate?...We are using RAID1 mirroring...Grecko
> >> >
> >> > "Jeff Pitsch" wrote:
> >> >
> >> >> 100 users does equal a lot fo traffic. Have you sniffed or
> >> >> analyzed the network traffic? is the link saturated? Have
> >> >> you looked at the queue for
> >> >> the hard drives to see if commands are lining up during peak
> >> >> periods? What
> >> >> RAID config are you using?
> >> >>
> >> >> Jeff Pitsch
> >> >> Microsoft MVP - Terminal Services
> >> >>
> >> >>
> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> >> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> >> >> > There is a lot of memory still available on the server
> >> >> > when the performance
> >> >> > is slow and no spikes in the processor. The only thing
> >> >> > that indicates a
> >> >> > problem is the high read/writes in the perfmon. Our
> >> >> > application people our
> >> >> > telling us that the bottle neck is in the network and that
> >> >> > the read/writes
> >> >> > shouldnt have an effect on performance. Of course, we are
> >> >> > seeing the opposite
> >> >> > of this. I am new to the environment and just trying to
> >> >> > help each side.
> >> >> > It
> >> >> > seems that it may be a little of both network and
> >> >> > hard-drive read/writes
> >> >> > as
> >> >> > best as I can tell. Any ideas?...Grecko
> >> >> >
> >> >> > "Jeff Pitsch" wrote:
> >> >> >
> >> >> >> I'm not sure how network would affect read/writes to the
> >> >> >> HD of the server.
> >> >> >> The only way it could, that I can think of off hand, is
> >> >> >> if the network card
> >> >> >> was bad and causing an unusual amount of interrupts
> >> >> >> forcing the processor
> >> >> >> to
> >> >> >> work harder and forcing more read/writes as the cpu is
> >> >> >> backed up more and
> >> >> >> more. What does the queue show for requests waiting to
> >> >> >> the hard drives?
> >> >> >> Is
> >> >> >> CPU being spiked at all? How much memory is available on
> >> >> >> the servers when
> >> >> >> this is happening?
> >> >> >>
> >> >> >> Jeff Pitsch
> >> >> >> Microsoft MVP - Terminal Services
> >> >> >>
> >> >> >>
> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> >> >> >> message
> >> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> >> >> >> > We havent tried anything yet. Yes, we are probably
> >> >> >> > going to add another
> >> >> >> > server to offload some of the users. We didnt want to
> >> >> >> > do this until we
> >> >> >> > had
> >> >> >> > figured out whats causing the problems.....network,
> >> >> >> > hard-drive, etc...Grecko
> >> >> >> >
> >> >> >> > "Jeff Pitsch" wrote:
> >> >> >> >
> >> >> >> >> Page files should also be min/max the same so windows
> >> >> >> >> is not growing/shrinking dynamically. this can add
> >> >> >> >> alot of overhead. Adding
> >> >> >> >> more
> >> >> >> >> pagefile space would just worsen the situation.
> >> >> >> >>
> >> >> >> >> both Vera and I have mentioned this but have you
> >> >> >> >> seriously considered
> >> >> >> >> that
> >> >> >> >> you may have simply maxed out the server? Have you
> >> >> >> >> tried virtual memory
> >> >> >> >> products from companies such as ThinPrint, RTO Soft,
> >> >> >> >> Provision Networks
> >> >> >> >> (I'm
> >> >> >> >> an employee)? Those would cut down on your paging
> >> >> >> >> drasticallyl if there
> >> >> >> >> are
> >> >> >> >> savings to be had.
> >> >> >> >>
> >> >> >> >> Jeff Pitsch
> >> >> >> >> Microsoft MVP - Terminal Services
> >> >> >> >>
> >> >> >> >>
> >> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> >> >> >> >> message
> >> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com
> >> >> >> >> ...
> >> >> >> >> >I checked the page file and its setup on drive C: for
> >> >> >> >> > 400 to 1000 MB,
> >> >> >> >> >on
> >> >> >> >> > drive F: for 12286 - 13386, drive H is a system
> >> >> >> >> > managed drive, and
> >> >> >> >> > driveE
> >> >> >> >> > has
> >> >> >> >> > no paging file. Should we make the amount lower on
> >> >> >> >> > C:. I am assuming
> >> >> >> >> > it
> >> >> >> >> > wont
> >> >> >> >> > roll to F until it uses the all page file on C:? Of
> >> >> >> >> > course if this
> >> >> >> >> > is
> >> >> >> >> > the
> >> >> >> >> > case, then we would have a bottle neck on the C:
> >> >> >> >> > drive. What would
> >> >> >> >> > be
> >> >> >> >> > the
> >> >> >> >> > best way to distribute the calls to the
> >> >> >> >> > drives?...Gregg
> >> >> >> >> >
> >> >> >> >> > "Vera Noest [MVP]" wrote:
> >> >> >> >> >
> >> >> >> >> >> It's impossible to say here why the server is
> >> >> >> >> >> having performance problems (but 100 concurrent
> >> >> >> >> >> sessions is quite a lot, depemding of
> >> >> >> >> >> course on which applications the users are
> >> >> >> >> >> running).
> >> >> >> >> >>
> >> >> >> >> >> You'll have to use Performance monitor to find out
> >> >> >> >> >> what the bottleneck is. Since you have noticed high
> >> >> >> >> >> RW activity on the drive,
> >> >> >> >> >> check the size and location of your swap file.
> >> >> >> >> >> You can find some tools and guidelines here:
> >> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
> >> >> >> >> >> ____________________________________________________
> >> >> >> >> >> _____ Vera Noest
> >> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> >> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
> >> >> >> >> >> ___ please respond in newsgroup, NOT by private
> >> >> >> >> >> email ___
> >> >> >> >> >>
> >> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
> >> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
> >> >> >> >> >> 03 jul 2008 in
> >> >> >> >> >> microsoft.public.windows.terminal_services:
> >> >> >> >> >>
> >> >> >> >> >> > We are having poor performance on our terminal
> >> >> >> >> >> > server. We only
> >> >> >> >> >> > have about 100 users on one server with 2
> >> >> >> >> >> > processors and 8 gigs
> >> >> >> >> >> > of ram. The poor performance seems to coincide
> >> >> >> >> >> > with high/read writes on our C: drive. Could the
> >> >> >> >> >> > drive itself be the bottleneck or the network
> >> >> >> >> >> > itself?...Thanks...Grecko

>
 
Re: Terminal Server Performance

After doing further research, it appears that the bottleneck is with the
pagefile. I have read conflicting reports that we should turn off the page
file entirely since we are not coming close to using all of our the physical
RAM or move it to another drive. Any thoughts on this?....Grecko

"Grecko" wrote:

> Thanks. I will try that out. So far today, the system is running smoothly.
> I will have to analyze the data when the server starts going slow again. Any
> other free programs you know of to monitor the traffic?...Grecko
>
> "Vera Noest [MVP]" wrote:
>
> > No, not from Task Manager. You'll need Network Monitor. Check if
> > this helps:
> >
> > How can I measure RDP bandwidth usage?
> > http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwidth
> >
> > _________________________________________________________
> > Vera Noest
> > MCSE, CCEA, Microsoft MVP - Terminal Server
> > TS troubleshooting: http://ts.veranoest.net
> > ___ please respond in newsgroup, NOT by private email ___
> >
> > =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> > 05 jul 2008 in microsoft.public.windows.terminal_services:
> >
> > > I hope you had a good 4th. I will have to look at the disk
> > > queue on Monday.
> > >
> > > We are using switches and the server has a 1 gbps card in it
> > > connecting it to the switch. I already told them they need to
> > > get a fiber connection from the server to the switch. Is there
> > > anyway to tell from the Windows Task Manager if the network
> > > connection is saturated? What is an acceptable range for the
> > > percentage in the network utilization?... I really appreciate
> > > the help...Grecko
> > >
> > > "Jeff Pitsch" wrote:
> > >
> > >> Read/Writes themselves aren't really indicitive of a problem
> > >> unless the queue is building up. If there is no queue then
> > >> your disks probably aren't a problem. In other words, no
> > >> commands are waiting for the disk/channel to free up.
> > >>
> > >> As for the tool, there's a bunch out there but the free ones
> > >> slip my mind. almost midnight and I'm tired :( but you really
> > >> are looking to see if the, I'm assuming, 100meg connection is
> > >> simply saturated tot he terminal server. Out of curiousity are
> > >> they using a hub or switch for all those single segment
> > >> connections?
> > >>
> > >> Jeff Pitsch
> > >> Microsoft MVP - Terminal Services
> > >>
> > >>
> > >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> > >> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
> > >> > No we havent sniffed or analyzed the network traffic. Like I
> > >> > said I know they have network issues because they still have
> > >> > static ips, there is no segments in the network, etc. What
> > >> > tool would you recommend to analyze the traffic? What do you
> > >> > mean by the link saturated? We havent looked at the queue
> > >> > yet during peak period...just the reads/writes. What would
> > >> > the queue
> > >> > indicate?...We are using RAID1 mirroring...Grecko
> > >> >
> > >> > "Jeff Pitsch" wrote:
> > >> >
> > >> >> 100 users does equal a lot fo traffic. Have you sniffed or
> > >> >> analyzed the network traffic? is the link saturated? Have
> > >> >> you looked at the queue for
> > >> >> the hard drives to see if commands are lining up during peak
> > >> >> periods? What
> > >> >> RAID config are you using?
> > >> >>
> > >> >> Jeff Pitsch
> > >> >> Microsoft MVP - Terminal Services
> > >> >>
> > >> >>
> > >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> > >> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> > >> >> > There is a lot of memory still available on the server
> > >> >> > when the performance
> > >> >> > is slow and no spikes in the processor. The only thing
> > >> >> > that indicates a
> > >> >> > problem is the high read/writes in the perfmon. Our
> > >> >> > application people our
> > >> >> > telling us that the bottle neck is in the network and that
> > >> >> > the read/writes
> > >> >> > shouldnt have an effect on performance. Of course, we are
> > >> >> > seeing the opposite
> > >> >> > of this. I am new to the environment and just trying to
> > >> >> > help each side.
> > >> >> > It
> > >> >> > seems that it may be a little of both network and
> > >> >> > hard-drive read/writes
> > >> >> > as
> > >> >> > best as I can tell. Any ideas?...Grecko
> > >> >> >
> > >> >> > "Jeff Pitsch" wrote:
> > >> >> >
> > >> >> >> I'm not sure how network would affect read/writes to the
> > >> >> >> HD of the server.
> > >> >> >> The only way it could, that I can think of off hand, is
> > >> >> >> if the network card
> > >> >> >> was bad and causing an unusual amount of interrupts
> > >> >> >> forcing the processor
> > >> >> >> to
> > >> >> >> work harder and forcing more read/writes as the cpu is
> > >> >> >> backed up more and
> > >> >> >> more. What does the queue show for requests waiting to
> > >> >> >> the hard drives?
> > >> >> >> Is
> > >> >> >> CPU being spiked at all? How much memory is available on
> > >> >> >> the servers when
> > >> >> >> this is happening?
> > >> >> >>
> > >> >> >> Jeff Pitsch
> > >> >> >> Microsoft MVP - Terminal Services
> > >> >> >>
> > >> >> >>
> > >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> > >> >> >> message
> > >> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> > >> >> >> > We havent tried anything yet. Yes, we are probably
> > >> >> >> > going to add another
> > >> >> >> > server to offload some of the users. We didnt want to
> > >> >> >> > do this until we
> > >> >> >> > had
> > >> >> >> > figured out whats causing the problems.....network,
> > >> >> >> > hard-drive, etc...Grecko
> > >> >> >> >
> > >> >> >> > "Jeff Pitsch" wrote:
> > >> >> >> >
> > >> >> >> >> Page files should also be min/max the same so windows
> > >> >> >> >> is not growing/shrinking dynamically. this can add
> > >> >> >> >> alot of overhead. Adding
> > >> >> >> >> more
> > >> >> >> >> pagefile space would just worsen the situation.
> > >> >> >> >>
> > >> >> >> >> both Vera and I have mentioned this but have you
> > >> >> >> >> seriously considered
> > >> >> >> >> that
> > >> >> >> >> you may have simply maxed out the server? Have you
> > >> >> >> >> tried virtual memory
> > >> >> >> >> products from companies such as ThinPrint, RTO Soft,
> > >> >> >> >> Provision Networks
> > >> >> >> >> (I'm
> > >> >> >> >> an employee)? Those would cut down on your paging
> > >> >> >> >> drasticallyl if there
> > >> >> >> >> are
> > >> >> >> >> savings to be had.
> > >> >> >> >>
> > >> >> >> >> Jeff Pitsch
> > >> >> >> >> Microsoft MVP - Terminal Services
> > >> >> >> >>
> > >> >> >> >>
> > >> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> > >> >> >> >> message
> > >> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com
> > >> >> >> >> ...
> > >> >> >> >> >I checked the page file and its setup on drive C: for
> > >> >> >> >> > 400 to 1000 MB,
> > >> >> >> >> >on
> > >> >> >> >> > drive F: for 12286 - 13386, drive H is a system
> > >> >> >> >> > managed drive, and
> > >> >> >> >> > driveE
> > >> >> >> >> > has
> > >> >> >> >> > no paging file. Should we make the amount lower on
> > >> >> >> >> > C:. I am assuming
> > >> >> >> >> > it
> > >> >> >> >> > wont
> > >> >> >> >> > roll to F until it uses the all page file on C:? Of
> > >> >> >> >> > course if this
> > >> >> >> >> > is
> > >> >> >> >> > the
> > >> >> >> >> > case, then we would have a bottle neck on the C:
> > >> >> >> >> > drive. What would
> > >> >> >> >> > be
> > >> >> >> >> > the
> > >> >> >> >> > best way to distribute the calls to the
> > >> >> >> >> > drives?...Gregg
> > >> >> >> >> >
> > >> >> >> >> > "Vera Noest [MVP]" wrote:
> > >> >> >> >> >
> > >> >> >> >> >> It's impossible to say here why the server is
> > >> >> >> >> >> having performance problems (but 100 concurrent
> > >> >> >> >> >> sessions is quite a lot, depemding of
> > >> >> >> >> >> course on which applications the users are
> > >> >> >> >> >> running).
> > >> >> >> >> >>
> > >> >> >> >> >> You'll have to use Performance monitor to find out
> > >> >> >> >> >> what the bottleneck is. Since you have noticed high
> > >> >> >> >> >> RW activity on the drive,
> > >> >> >> >> >> check the size and location of your swap file.
> > >> >> >> >> >> You can find some tools and guidelines here:
> > >> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
> > >> >> >> >> >> ____________________________________________________
> > >> >> >> >> >> _____ Vera Noest
> > >> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> > >> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
> > >> >> >> >> >> ___ please respond in newsgroup, NOT by private
> > >> >> >> >> >> email ___
> > >> >> >> >> >>
> > >> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
> > >> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
> > >> >> >> >> >> 03 jul 2008 in
> > >> >> >> >> >> microsoft.public.windows.terminal_services:
> > >> >> >> >> >>
> > >> >> >> >> >> > We are having poor performance on our terminal
> > >> >> >> >> >> > server. We only
> > >> >> >> >> >> > have about 100 users on one server with 2
> > >> >> >> >> >> > processors and 8 gigs
> > >> >> >> >> >> > of ram. The poor performance seems to coincide
> > >> >> >> >> >> > with high/read writes on our C: drive. Could the
> > >> >> >> >> >> > drive itself be the bottleneck or the network
> > >> >> >> >> >> > itself?...Thanks...Grecko

> >
 
Re: Terminal Server Performance

I would definitively move the page file away from the C drive. Make
it a fixed amount (same low and high size) of 1,5 - 2 times the
amount of physical RAM on another disk, if possible with another disk
controller.

How to determine the appropriate page file size for 64-bit versions
of Windows Server 2003 or Windows XP
http://support.microsoft.com/kb/889654/en-us
_________________________________________________________
Vera Noest
MCSE, CCEA, Microsoft MVP - Terminal Server
TS troubleshooting: http://ts.veranoest.net
___ please respond in newsgroup, NOT by private email ___

=?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
09 jul 2008 in microsoft.public.windows.terminal_services:

> After doing further research, it appears that the bottleneck is
> with the pagefile. I have read conflicting reports that we
> should turn off the page file entirely since we are not coming
> close to using all of our the physical RAM or move it to another
> drive. Any thoughts on this?....Grecko
>
> "Grecko" wrote:
>
>> Thanks. I will try that out. So far today, the system is
>> running smoothly. I will have to analyze the data when the
>> server starts going slow again. Any other free programs you
>> know of to monitor the traffic?...Grecko
>>
>> "Vera Noest [MVP]" wrote:
>>
>> > No, not from Task Manager. You'll need Network Monitor. Check
>> > if this helps:
>> >
>> > How can I measure RDP bandwidth usage?
>> > http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwid
>> > th
>> >
>> > _________________________________________________________
>> > Vera Noest
>> > MCSE, CCEA, Microsoft MVP - Terminal Server
>> > TS troubleshooting: http://ts.veranoest.net
>> > ___ please respond in newsgroup, NOT by private email ___
>> >
>> > =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote
>> > on 05 jul 2008 in microsoft.public.windows.terminal_services:
>> >
>> > > I hope you had a good 4th. I will have to look at the disk
>> > > queue on Monday.
>> > >
>> > > We are using switches and the server has a 1 gbps card in
>> > > it connecting it to the switch. I already told them they
>> > > need to get a fiber connection from the server to the
>> > > switch. Is there anyway to tell from the Windows Task
>> > > Manager if the network connection is saturated? What is an
>> > > acceptable range for the percentage in the network
>> > > utilization?... I really appreciate the help...Grecko
>> > >
>> > > "Jeff Pitsch" wrote:
>> > >
>> > >> Read/Writes themselves aren't really indicitive of a
>> > >> problem unless the queue is building up. If there is no
>> > >> queue then your disks probably aren't a problem. In other
>> > >> words, no commands are waiting for the disk/channel to
>> > >> free up.
>> > >>
>> > >> As for the tool, there's a bunch out there but the free
>> > >> ones slip my mind. almost midnight and I'm tired :( but
>> > >> you really are looking to see if the, I'm assuming, 100meg
>> > >> connection is simply saturated tot he terminal server. Out
>> > >> of curiousity are they using a hub or switch for all those
>> > >> single segment connections?
>> > >>
>> > >> Jeff Pitsch
>> > >> Microsoft MVP - Terminal Services
>> > >>
>> > >>
>> > >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> > >> message
>> > >> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
>> > >> > No we havent sniffed or analyzed the network traffic.
>> > >> > Like I said I know they have network issues because they
>> > >> > still have static ips, there is no segments in the
>> > >> > network, etc. What tool would you recommend to analyze
>> > >> > the traffic? What do you mean by the link saturated? We
>> > >> > havent looked at the queue yet during peak period...just
>> > >> > the reads/writes. What would the queue
>> > >> > indicate?...We are using RAID1 mirroring...Grecko
>> > >> >
>> > >> > "Jeff Pitsch" wrote:
>> > >> >
>> > >> >> 100 users does equal a lot fo traffic. Have you
>> > >> >> sniffed or analyzed the network traffic? is the link
>> > >> >> saturated? Have you looked at the queue for
>> > >> >> the hard drives to see if commands are lining up during
>> > >> >> peak periods? What
>> > >> >> RAID config are you using?
>> > >> >>
>> > >> >> Jeff Pitsch
>> > >> >> Microsoft MVP - Terminal Services
>> > >> >>
>> > >> >>
>> > >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> > >> >> message
>> > >> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com.
>> > >> >> ..
>> > >> >> > There is a lot of memory still available on the
>> > >> >> > server when the performance
>> > >> >> > is slow and no spikes in the processor. The only
>> > >> >> > thing that indicates a
>> > >> >> > problem is the high read/writes in the perfmon. Our
>> > >> >> > application people our
>> > >> >> > telling us that the bottle neck is in the network and
>> > >> >> > that the read/writes
>> > >> >> > shouldnt have an effect on performance. Of course, we
>> > >> >> > are seeing the opposite
>> > >> >> > of this. I am new to the environment and just trying
>> > >> >> > to help each side.
>> > >> >> > It
>> > >> >> > seems that it may be a little of both network and
>> > >> >> > hard-drive read/writes
>> > >> >> > as
>> > >> >> > best as I can tell. Any ideas?...Grecko
>> > >> >> >
>> > >> >> > "Jeff Pitsch" wrote:
>> > >> >> >
>> > >> >> >> I'm not sure how network would affect read/writes to
>> > >> >> >> the HD of the server.
>> > >> >> >> The only way it could, that I can think of off hand,
>> > >> >> >> is if the network card
>> > >> >> >> was bad and causing an unusual amount of interrupts
>> > >> >> >> forcing the processor
>> > >> >> >> to
>> > >> >> >> work harder and forcing more read/writes as the cpu
>> > >> >> >> is backed up more and
>> > >> >> >> more. What does the queue show for requests waiting
>> > >> >> >> to the hard drives?
>> > >> >> >> Is
>> > >> >> >> CPU being spiked at all? How much memory is
>> > >> >> >> available on the servers when
>> > >> >> >> this is happening?
>> > >> >> >>
>> > >> >> >> Jeff Pitsch
>> > >> >> >> Microsoft MVP - Terminal Services
>> > >> >> >>
>> > >> >> >>
>> > >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> > >> >> >> message
>> > >> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.c
>> > >> >> >> om...
>> > >> >> >> > We havent tried anything yet. Yes, we are
>> > >> >> >> > probably going to add another
>> > >> >> >> > server to offload some of the users. We didnt
>> > >> >> >> > want to do this until we
>> > >> >> >> > had
>> > >> >> >> > figured out whats causing the
>> > >> >> >> > problems.....network, hard-drive, etc...Grecko
>> > >> >> >> >
>> > >> >> >> > "Jeff Pitsch" wrote:
>> > >> >> >> >
>> > >> >> >> >> Page files should also be min/max the same so
>> > >> >> >> >> windows is not growing/shrinking dynamically.
>> > >> >> >> >> this can add alot of overhead. Adding
>> > >> >> >> >> more
>> > >> >> >> >> pagefile space would just worsen the situation.
>> > >> >> >> >>
>> > >> >> >> >> both Vera and I have mentioned this but have you
>> > >> >> >> >> seriously considered
>> > >> >> >> >> that
>> > >> >> >> >> you may have simply maxed out the server? Have
>> > >> >> >> >> you tried virtual memory
>> > >> >> >> >> products from companies such as ThinPrint, RTO
>> > >> >> >> >> Soft, Provision Networks
>> > >> >> >> >> (I'm
>> > >> >> >> >> an employee)? Those would cut down on your
>> > >> >> >> >> paging drasticallyl if there
>> > >> >> >> >> are
>> > >> >> >> >> savings to be had.
>> > >> >> >> >>
>> > >> >> >> >> Jeff Pitsch
>> > >> >> >> >> Microsoft MVP - Terminal Services
>> > >> >> >> >>
>> > >> >> >> >>
>> > >> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote
>> > >> >> >> >> in message
>> > >> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsof
>> > >> >> >> >> t.com ...
>> > >> >> >> >> >I checked the page file and its setup on drive
>> > >> >> >> >> >C: for
>> > >> >> >> >> > 400 to 1000 MB,
>> > >> >> >> >> >on
>> > >> >> >> >> > drive F: for 12286 - 13386, drive H is a system
>> > >> >> >> >> > managed drive, and
>> > >> >> >> >> > driveE
>> > >> >> >> >> > has
>> > >> >> >> >> > no paging file. Should we make the amount
>> > >> >> >> >> > lower on C:. I am assuming
>> > >> >> >> >> > it
>> > >> >> >> >> > wont
>> > >> >> >> >> > roll to F until it uses the all page file on
>> > >> >> >> >> > C:? Of course if this
>> > >> >> >> >> > is
>> > >> >> >> >> > the
>> > >> >> >> >> > case, then we would have a bottle neck on the
>> > >> >> >> >> > C: drive. What would
>> > >> >> >> >> > be
>> > >> >> >> >> > the
>> > >> >> >> >> > best way to distribute the calls to the
>> > >> >> >> >> > drives?...Gregg
>> > >> >> >> >> >
>> > >> >> >> >> > "Vera Noest [MVP]" wrote:
>> > >> >> >> >> >
>> > >> >> >> >> >> It's impossible to say here why the server is
>> > >> >> >> >> >> having performance problems (but 100
>> > >> >> >> >> >> concurrent sessions is quite a lot, depemding
>> > >> >> >> >> >> of course on which applications the users are
>> > >> >> >> >> >> running).
>> > >> >> >> >> >>
>> > >> >> >> >> >> You'll have to use Performance monitor to find
>> > >> >> >> >> >> out what the bottleneck is. Since you have
>> > >> >> >> >> >> noticed high RW activity on the drive,
>> > >> >> >> >> >> check the size and location of your swap file.
>> > >> >> >> >> >> You can find some tools and guidelines here:
>> > >> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
>> > >> >> >> >> >> _______________________________________________
>> > >> >> >> >> >> _____ _____ Vera Noest
>> > >> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> > >> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
>> > >> >> >> >> >> ___ please respond in newsgroup, NOT by
>> > >> >> >> >> >> private email ___
>> > >> >> >> >> >>
>> > >> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
>> > >> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
>> > >> >> >> >> >> 03 jul 2008 in
>> > >> >> >> >> >> microsoft.public.windows.terminal_services:
>> > >> >> >> >> >>
>> > >> >> >> >> >> > We are having poor performance on our
>> > >> >> >> >> >> > terminal server. We only
>> > >> >> >> >> >> > have about 100 users on one server with 2
>> > >> >> >> >> >> > processors and 8 gigs
>> > >> >> >> >> >> > of ram. The poor performance seems to
>> > >> >> >> >> >> > coincide with high/read writes on our C:
>> > >> >> >> >> >> > drive. Could the drive itself be the
>> > >> >> >> >> >> > bottleneck or the network
>> > >> >> >> >> >> > itself?...Thanks...Grecko
 
Re: Terminal Server Performance

After further review, we are finding the disk queue is in fact building up
and we are still seeing problems with performance. I have run network tests
when we are having problems and it seems the network is fine during the
problem periods as well as the CPU and memory utilization. The only logical
conclusion then is that the bottleneck is with the C drive. Of course the
problems dont happen in a pattern so its hard to trace the problem back to
its origin. What suggestions would you have to fix this?....Gregg

"Grecko" wrote:

> Thanks. I will try that out. So far today, the system is running smoothly.
> I will have to analyze the data when the server starts going slow again. Any
> other free programs you know of to monitor the traffic?...Grecko
>
> "Vera Noest [MVP]" wrote:
>
> > No, not from Task Manager. You'll need Network Monitor. Check if
> > this helps:
> >
> > How can I measure RDP bandwidth usage?
> > http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwidth
> >
> > _________________________________________________________
> > Vera Noest
> > MCSE, CCEA, Microsoft MVP - Terminal Server
> > TS troubleshooting: http://ts.veranoest.net
> > ___ please respond in newsgroup, NOT by private email ___
> >
> > =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
> > 05 jul 2008 in microsoft.public.windows.terminal_services:
> >
> > > I hope you had a good 4th. I will have to look at the disk
> > > queue on Monday.
> > >
> > > We are using switches and the server has a 1 gbps card in it
> > > connecting it to the switch. I already told them they need to
> > > get a fiber connection from the server to the switch. Is there
> > > anyway to tell from the Windows Task Manager if the network
> > > connection is saturated? What is an acceptable range for the
> > > percentage in the network utilization?... I really appreciate
> > > the help...Grecko
> > >
> > > "Jeff Pitsch" wrote:
> > >
> > >> Read/Writes themselves aren't really indicitive of a problem
> > >> unless the queue is building up. If there is no queue then
> > >> your disks probably aren't a problem. In other words, no
> > >> commands are waiting for the disk/channel to free up.
> > >>
> > >> As for the tool, there's a bunch out there but the free ones
> > >> slip my mind. almost midnight and I'm tired :( but you really
> > >> are looking to see if the, I'm assuming, 100meg connection is
> > >> simply saturated tot he terminal server. Out of curiousity are
> > >> they using a hub or switch for all those single segment
> > >> connections?
> > >>
> > >> Jeff Pitsch
> > >> Microsoft MVP - Terminal Services
> > >>
> > >>
> > >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> > >> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
> > >> > No we havent sniffed or analyzed the network traffic. Like I
> > >> > said I know they have network issues because they still have
> > >> > static ips, there is no segments in the network, etc. What
> > >> > tool would you recommend to analyze the traffic? What do you
> > >> > mean by the link saturated? We havent looked at the queue
> > >> > yet during peak period...just the reads/writes. What would
> > >> > the queue
> > >> > indicate?...We are using RAID1 mirroring...Grecko
> > >> >
> > >> > "Jeff Pitsch" wrote:
> > >> >
> > >> >> 100 users does equal a lot fo traffic. Have you sniffed or
> > >> >> analyzed the network traffic? is the link saturated? Have
> > >> >> you looked at the queue for
> > >> >> the hard drives to see if commands are lining up during peak
> > >> >> periods? What
> > >> >> RAID config are you using?
> > >> >>
> > >> >> Jeff Pitsch
> > >> >> Microsoft MVP - Terminal Services
> > >> >>
> > >> >>
> > >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
> > >> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
> > >> >> > There is a lot of memory still available on the server
> > >> >> > when the performance
> > >> >> > is slow and no spikes in the processor. The only thing
> > >> >> > that indicates a
> > >> >> > problem is the high read/writes in the perfmon. Our
> > >> >> > application people our
> > >> >> > telling us that the bottle neck is in the network and that
> > >> >> > the read/writes
> > >> >> > shouldnt have an effect on performance. Of course, we are
> > >> >> > seeing the opposite
> > >> >> > of this. I am new to the environment and just trying to
> > >> >> > help each side.
> > >> >> > It
> > >> >> > seems that it may be a little of both network and
> > >> >> > hard-drive read/writes
> > >> >> > as
> > >> >> > best as I can tell. Any ideas?...Grecko
> > >> >> >
> > >> >> > "Jeff Pitsch" wrote:
> > >> >> >
> > >> >> >> I'm not sure how network would affect read/writes to the
> > >> >> >> HD of the server.
> > >> >> >> The only way it could, that I can think of off hand, is
> > >> >> >> if the network card
> > >> >> >> was bad and causing an unusual amount of interrupts
> > >> >> >> forcing the processor
> > >> >> >> to
> > >> >> >> work harder and forcing more read/writes as the cpu is
> > >> >> >> backed up more and
> > >> >> >> more. What does the queue show for requests waiting to
> > >> >> >> the hard drives?
> > >> >> >> Is
> > >> >> >> CPU being spiked at all? How much memory is available on
> > >> >> >> the servers when
> > >> >> >> this is happening?
> > >> >> >>
> > >> >> >> Jeff Pitsch
> > >> >> >> Microsoft MVP - Terminal Services
> > >> >> >>
> > >> >> >>
> > >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> > >> >> >> message
> > >> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
> > >> >> >> > We havent tried anything yet. Yes, we are probably
> > >> >> >> > going to add another
> > >> >> >> > server to offload some of the users. We didnt want to
> > >> >> >> > do this until we
> > >> >> >> > had
> > >> >> >> > figured out whats causing the problems.....network,
> > >> >> >> > hard-drive, etc...Grecko
> > >> >> >> >
> > >> >> >> > "Jeff Pitsch" wrote:
> > >> >> >> >
> > >> >> >> >> Page files should also be min/max the same so windows
> > >> >> >> >> is not growing/shrinking dynamically. this can add
> > >> >> >> >> alot of overhead. Adding
> > >> >> >> >> more
> > >> >> >> >> pagefile space would just worsen the situation.
> > >> >> >> >>
> > >> >> >> >> both Vera and I have mentioned this but have you
> > >> >> >> >> seriously considered
> > >> >> >> >> that
> > >> >> >> >> you may have simply maxed out the server? Have you
> > >> >> >> >> tried virtual memory
> > >> >> >> >> products from companies such as ThinPrint, RTO Soft,
> > >> >> >> >> Provision Networks
> > >> >> >> >> (I'm
> > >> >> >> >> an employee)? Those would cut down on your paging
> > >> >> >> >> drasticallyl if there
> > >> >> >> >> are
> > >> >> >> >> savings to be had.
> > >> >> >> >>
> > >> >> >> >> Jeff Pitsch
> > >> >> >> >> Microsoft MVP - Terminal Services
> > >> >> >> >>
> > >> >> >> >>
> > >> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
> > >> >> >> >> message
> > >> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com
> > >> >> >> >> ...
> > >> >> >> >> >I checked the page file and its setup on drive C: for
> > >> >> >> >> > 400 to 1000 MB,
> > >> >> >> >> >on
> > >> >> >> >> > drive F: for 12286 - 13386, drive H is a system
> > >> >> >> >> > managed drive, and
> > >> >> >> >> > driveE
> > >> >> >> >> > has
> > >> >> >> >> > no paging file. Should we make the amount lower on
> > >> >> >> >> > C:. I am assuming
> > >> >> >> >> > it
> > >> >> >> >> > wont
> > >> >> >> >> > roll to F until it uses the all page file on C:? Of
> > >> >> >> >> > course if this
> > >> >> >> >> > is
> > >> >> >> >> > the
> > >> >> >> >> > case, then we would have a bottle neck on the C:
> > >> >> >> >> > drive. What would
> > >> >> >> >> > be
> > >> >> >> >> > the
> > >> >> >> >> > best way to distribute the calls to the
> > >> >> >> >> > drives?...Gregg
> > >> >> >> >> >
> > >> >> >> >> > "Vera Noest [MVP]" wrote:
> > >> >> >> >> >
> > >> >> >> >> >> It's impossible to say here why the server is
> > >> >> >> >> >> having performance problems (but 100 concurrent
> > >> >> >> >> >> sessions is quite a lot, depemding of
> > >> >> >> >> >> course on which applications the users are
> > >> >> >> >> >> running).
> > >> >> >> >> >>
> > >> >> >> >> >> You'll have to use Performance monitor to find out
> > >> >> >> >> >> what the bottleneck is. Since you have noticed high
> > >> >> >> >> >> RW activity on the drive,
> > >> >> >> >> >> check the size and location of your swap file.
> > >> >> >> >> >> You can find some tools and guidelines here:
> > >> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
> > >> >> >> >> >> ____________________________________________________
> > >> >> >> >> >> _____ Vera Noest
> > >> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
> > >> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
> > >> >> >> >> >> ___ please respond in newsgroup, NOT by private
> > >> >> >> >> >> email ___
> > >> >> >> >> >>
> > >> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
> > >> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
> > >> >> >> >> >> 03 jul 2008 in
> > >> >> >> >> >> microsoft.public.windows.terminal_services:
> > >> >> >> >> >>
> > >> >> >> >> >> > We are having poor performance on our terminal
> > >> >> >> >> >> > server. We only
> > >> >> >> >> >> > have about 100 users on one server with 2
> > >> >> >> >> >> > processors and 8 gigs
> > >> >> >> >> >> > of ram. The poor performance seems to coincide
> > >> >> >> >> >> > with high/read writes on our C: drive. Could the
> > >> >> >> >> >> > drive itself be the bottleneck or the network
> > >> >> >> >> >> > itself?...Thanks...Grecko

> >
 
Re: Terminal Server Performance

You may want to try a virtual memory manager like something from RTOSoft's
TScale product. This would help cut down on page faults. Moving the page
file to another disk and, more importantly, disk controller. If you want to
be really fast, move it to Solid State drive. Amazing how much of a
difference that makes. If it's an application causing the queue to build
up, then you may want to consider moving the application as well.

Jeff Pitsch
Microsoft MVP - Terminal Services


"Grecko" <Grecko@discussions.microsoft.com> wrote in message
news:0DB051E4-082E-46A6-9E9A-40D957ABF20A@microsoft.com...
> After further review, we are finding the disk queue is in fact building up
> and we are still seeing problems with performance. I have run network
> tests
> when we are having problems and it seems the network is fine during the
> problem periods as well as the CPU and memory utilization. The only
> logical
> conclusion then is that the bottleneck is with the C drive. Of course the
> problems dont happen in a pattern so its hard to trace the problem back to
> its origin. What suggestions would you have to fix this?....Gregg
>
> "Grecko" wrote:
>
>> Thanks. I will try that out. So far today, the system is running
>> smoothly.
>> I will have to analyze the data when the server starts going slow again.
>> Any
>> other free programs you know of to monitor the traffic?...Grecko
>>
>> "Vera Noest [MVP]" wrote:
>>
>> > No, not from Task Manager. You'll need Network Monitor. Check if
>> > this helps:
>> >
>> > How can I measure RDP bandwidth usage?
>> > http://ts.veranoest.net/ts_faq_performance.htm#monitor_bandwidth
>> >
>> > _________________________________________________________
>> > Vera Noest
>> > MCSE, CCEA, Microsoft MVP - Terminal Server
>> > TS troubleshooting: http://ts.veranoest.net
>> > ___ please respond in newsgroup, NOT by private email ___
>> >
>> > =?Utf-8?B?R3JlY2tv?= <Grecko@discussions.microsoft.com> wrote on
>> > 05 jul 2008 in microsoft.public.windows.terminal_services:
>> >
>> > > I hope you had a good 4th. I will have to look at the disk
>> > > queue on Monday.
>> > >
>> > > We are using switches and the server has a 1 gbps card in it
>> > > connecting it to the switch. I already told them they need to
>> > > get a fiber connection from the server to the switch. Is there
>> > > anyway to tell from the Windows Task Manager if the network
>> > > connection is saturated? What is an acceptable range for the
>> > > percentage in the network utilization?... I really appreciate
>> > > the help...Grecko
>> > >
>> > > "Jeff Pitsch" wrote:
>> > >
>> > >> Read/Writes themselves aren't really indicitive of a problem
>> > >> unless the queue is building up. If there is no queue then
>> > >> your disks probably aren't a problem. In other words, no
>> > >> commands are waiting for the disk/channel to free up.
>> > >>
>> > >> As for the tool, there's a bunch out there but the free ones
>> > >> slip my mind. almost midnight and I'm tired :( but you really
>> > >> are looking to see if the, I'm assuming, 100meg connection is
>> > >> simply saturated tot he terminal server. Out of curiousity are
>> > >> they using a hub or switch for all those single segment
>> > >> connections?
>> > >>
>> > >> Jeff Pitsch
>> > >> Microsoft MVP - Terminal Services
>> > >>
>> > >>
>> > >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> > >> news:D0791121-8F28-46BA-BE83-E9AFC47696D6@microsoft.com...
>> > >> > No we havent sniffed or analyzed the network traffic. Like I
>> > >> > said I know they have network issues because they still have
>> > >> > static ips, there is no segments in the network, etc. What
>> > >> > tool would you recommend to analyze the traffic? What do you
>> > >> > mean by the link saturated? We havent looked at the queue
>> > >> > yet during peak period...just the reads/writes. What would
>> > >> > the queue
>> > >> > indicate?...We are using RAID1 mirroring...Grecko
>> > >> >
>> > >> > "Jeff Pitsch" wrote:
>> > >> >
>> > >> >> 100 users does equal a lot fo traffic. Have you sniffed or
>> > >> >> analyzed the network traffic? is the link saturated? Have
>> > >> >> you looked at the queue for
>> > >> >> the hard drives to see if commands are lining up during peak
>> > >> >> periods? What
>> > >> >> RAID config are you using?
>> > >> >>
>> > >> >> Jeff Pitsch
>> > >> >> Microsoft MVP - Terminal Services
>> > >> >>
>> > >> >>
>> > >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in message
>> > >> >> news:55B7AA77-B11B-48AA-A66D-B060F6D2808E@microsoft.com...
>> > >> >> > There is a lot of memory still available on the server
>> > >> >> > when the performance
>> > >> >> > is slow and no spikes in the processor. The only thing
>> > >> >> > that indicates a
>> > >> >> > problem is the high read/writes in the perfmon. Our
>> > >> >> > application people our
>> > >> >> > telling us that the bottle neck is in the network and that
>> > >> >> > the read/writes
>> > >> >> > shouldnt have an effect on performance. Of course, we are
>> > >> >> > seeing the opposite
>> > >> >> > of this. I am new to the environment and just trying to
>> > >> >> > help each side.
>> > >> >> > It
>> > >> >> > seems that it may be a little of both network and
>> > >> >> > hard-drive read/writes
>> > >> >> > as
>> > >> >> > best as I can tell. Any ideas?...Grecko
>> > >> >> >
>> > >> >> > "Jeff Pitsch" wrote:
>> > >> >> >
>> > >> >> >> I'm not sure how network would affect read/writes to the
>> > >> >> >> HD of the server.
>> > >> >> >> The only way it could, that I can think of off hand, is
>> > >> >> >> if the network card
>> > >> >> >> was bad and causing an unusual amount of interrupts
>> > >> >> >> forcing the processor
>> > >> >> >> to
>> > >> >> >> work harder and forcing more read/writes as the cpu is
>> > >> >> >> backed up more and
>> > >> >> >> more. What does the queue show for requests waiting to
>> > >> >> >> the hard drives?
>> > >> >> >> Is
>> > >> >> >> CPU being spiked at all? How much memory is available on
>> > >> >> >> the servers when
>> > >> >> >> this is happening?
>> > >> >> >>
>> > >> >> >> Jeff Pitsch
>> > >> >> >> Microsoft MVP - Terminal Services
>> > >> >> >>
>> > >> >> >>
>> > >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> > >> >> >> message
>> > >> >> >> news:C6ECF95E-4FA8-4034-889E-D1C2E28DCB6D@microsoft.com...
>> > >> >> >> > We havent tried anything yet. Yes, we are probably
>> > >> >> >> > going to add another
>> > >> >> >> > server to offload some of the users. We didnt want to
>> > >> >> >> > do this until we
>> > >> >> >> > had
>> > >> >> >> > figured out whats causing the problems.....network,
>> > >> >> >> > hard-drive, etc...Grecko
>> > >> >> >> >
>> > >> >> >> > "Jeff Pitsch" wrote:
>> > >> >> >> >
>> > >> >> >> >> Page files should also be min/max the same so windows
>> > >> >> >> >> is not growing/shrinking dynamically. this can add
>> > >> >> >> >> alot of overhead. Adding
>> > >> >> >> >> more
>> > >> >> >> >> pagefile space would just worsen the situation.
>> > >> >> >> >>
>> > >> >> >> >> both Vera and I have mentioned this but have you
>> > >> >> >> >> seriously considered
>> > >> >> >> >> that
>> > >> >> >> >> you may have simply maxed out the server? Have you
>> > >> >> >> >> tried virtual memory
>> > >> >> >> >> products from companies such as ThinPrint, RTO Soft,
>> > >> >> >> >> Provision Networks
>> > >> >> >> >> (I'm
>> > >> >> >> >> an employee)? Those would cut down on your paging
>> > >> >> >> >> drasticallyl if there
>> > >> >> >> >> are
>> > >> >> >> >> savings to be had.
>> > >> >> >> >>
>> > >> >> >> >> Jeff Pitsch
>> > >> >> >> >> Microsoft MVP - Terminal Services
>> > >> >> >> >>
>> > >> >> >> >>
>> > >> >> >> >> "Grecko" <Grecko@discussions.microsoft.com> wrote in
>> > >> >> >> >> message
>> > >> >> >> >> news:6E96045F-ACCE-4752-8113-2F175A1FF2A7@microsoft.com
>> > >> >> >> >> ...
>> > >> >> >> >> >I checked the page file and its setup on drive C: for
>> > >> >> >> >> > 400 to 1000 MB,
>> > >> >> >> >> >on
>> > >> >> >> >> > drive F: for 12286 - 13386, drive H is a system
>> > >> >> >> >> > managed drive, and
>> > >> >> >> >> > driveE
>> > >> >> >> >> > has
>> > >> >> >> >> > no paging file. Should we make the amount lower on
>> > >> >> >> >> > C:. I am assuming
>> > >> >> >> >> > it
>> > >> >> >> >> > wont
>> > >> >> >> >> > roll to F until it uses the all page file on C:? Of
>> > >> >> >> >> > course if this
>> > >> >> >> >> > is
>> > >> >> >> >> > the
>> > >> >> >> >> > case, then we would have a bottle neck on the C:
>> > >> >> >> >> > drive. What would
>> > >> >> >> >> > be
>> > >> >> >> >> > the
>> > >> >> >> >> > best way to distribute the calls to the
>> > >> >> >> >> > drives?...Gregg
>> > >> >> >> >> >
>> > >> >> >> >> > "Vera Noest [MVP]" wrote:
>> > >> >> >> >> >
>> > >> >> >> >> >> It's impossible to say here why the server is
>> > >> >> >> >> >> having performance problems (but 100 concurrent
>> > >> >> >> >> >> sessions is quite a lot, depemding of
>> > >> >> >> >> >> course on which applications the users are
>> > >> >> >> >> >> running).
>> > >> >> >> >> >>
>> > >> >> >> >> >> You'll have to use Performance monitor to find out
>> > >> >> >> >> >> what the bottleneck is. Since you have noticed high
>> > >> >> >> >> >> RW activity on the drive,
>> > >> >> >> >> >> check the size and location of your swap file.
>> > >> >> >> >> >> You can find some tools and guidelines here:
>> > >> >> >> >> >> http://ts.veranoest.net/ts_performance.htm
>> > >> >> >> >> >> ____________________________________________________
>> > >> >> >> >> >> _____ Vera Noest
>> > >> >> >> >> >> MCSE, CCEA, Microsoft MVP - Terminal Server
>> > >> >> >> >> >> TS troubleshooting: http://ts.veranoest.net
>> > >> >> >> >> >> ___ please respond in newsgroup, NOT by private
>> > >> >> >> >> >> email ___
>> > >> >> >> >> >>
>> > >> >> >> >> >> =?Utf-8?B?R3JlY2tv?=
>> > >> >> >> >> >> <Grecko@discussions.microsoft.com> wrote on
>> > >> >> >> >> >> 03 jul 2008 in
>> > >> >> >> >> >> microsoft.public.windows.terminal_services:
>> > >> >> >> >> >>
>> > >> >> >> >> >> > We are having poor performance on our terminal
>> > >> >> >> >> >> > server. We only
>> > >> >> >> >> >> > have about 100 users on one server with 2
>> > >> >> >> >> >> > processors and 8 gigs
>> > >> >> >> >> >> > of ram. The poor performance seems to coincide
>> > >> >> >> >> >> > with high/read writes on our C: drive. Could the
>> > >> >> >> >> >> > drive itself be the bottleneck or the network
>> > >> >> >> >> >> > itself?...Thanks...Grecko
>> >
 
Back
Top