carillian
asked on
terminal services -memory allocation
in the real world how much server memory generally is required for a typical terminal services session?
any good links for examples on this?
any good links for examples on this?
128 per user???
We use a TS with 4GB to service 100 concurrent connections over Citrix!! Depends what your doing of course, you need to test to be sure but we don't give any where near 128 per user.
We use a TS with 4GB to service 100 concurrent connections over Citrix!! Depends what your doing of course, you need to test to be sure but we don't give any where near 128 per user.
Hi youre1m,
I don't think that that will speed up very well. We did test with the office suite an two of our core apps and needed at and about 100 Mb per user. So i would say 128 is a safe bet. If you use 40 Mb as you do, than i do believe users will start complaning about speed (and it's better to have 128 at first so they don't start wining around).
Also, you use Citrix which, i do believe, uses a bit less memory and bandwidth...
I don't think that that will speed up very well. We did test with the office suite an two of our core apps and needed at and about 100 Mb per user. So i would say 128 is a safe bet. If you use 40 Mb as you do, than i do believe users will start complaning about speed (and it's better to have 128 at first so they don't start wining around).
Also, you use Citrix which, i do believe, uses a bit less memory and bandwidth...
ASKER
thanks for input, 128 per user seems a lot also, I saw some microsoft calcs which said 128 + 4/5mb per user which 1) i couldn't belive and 2) don't know where I saw it.
this is not ctirix, so are we good looking at between 64 and 128 per user?
given a particular app, (in this instance a financial package) how I can I measure the amount of memory required?
this is not ctirix, so are we good looking at between 64 and 128 per user?
given a particular app, (in this instance a financial package) how I can I measure the amount of memory required?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
You have to test, there's no other way. Some apps are memory intensive, others not. The best bet is to follow the above suggestion. Benchmark the server with no users connected, connect 1 user, monitor memeory usage while they use the app. Now connect 5 users, do the same again, connect 10 users, same again. Reason being that you need to also see whether the memory usage goes up at an equal rate for as the user numbers increase.
eg. You may find that 1 user increases memory usage by 20mb, then 5 users 110mb, 10 users 300mb. It's a matter of how much time and resources you have to make the test as realistic as possible. Ideally you would use loadrunner or something similar to perform an action several times over to keep the tests consistent. It also means you don;t have to have 10 people to help you perform a load test.
eg. You may find that 1 user increases memory usage by 20mb, then 5 users 110mb, 10 users 300mb. It's a matter of how much time and resources you have to make the test as realistic as possible. Ideally you would use loadrunner or something similar to perform an action several times over to keep the tests consistent. It also means you don;t have to have 10 people to help you perform a load test.
ASKER
found the resource!
Have a look at http://www.microsoft.com/downloads/details.aspx?FamilyID=a2ae95da-be56-4495-9fb5-e4b7170b33d9&DisplayLang=en
and then at the section on "system and user memory requirements"
It seems to say that a "Data Entry Worker" using Microsoft Excel requires around 3.5MB per user!
How can they state this if not true?
Have a look at http://www.microsoft.com/downloads/details.aspx?FamilyID=a2ae95da-be56-4495-9fb5-e4b7170b33d9&DisplayLang=en
and then at the section on "system and user memory requirements"
It seems to say that a "Data Entry Worker" using Microsoft Excel requires around 3.5MB per user!
How can they state this if not true?
Hi,
Could be that they are right, but normally if Microsoft says that you only need 128 mb (minimum) to let XP work, it wouldn't work either. Normally, if microsoft says something about memory usage, i always multiply it with 2. But indeed i could be wrong about the 64-128 Mb. Maybe your core software uses more memory. Problem with TS is (before going into a to long discussion ;)) is that you need to test it. Maybe 30 or 40 MB is enough for one session, maybe 10 Mb is enough (which i don't think) or maybe you do need 128 Mb.
TS 2003 is much better than TS 2000, maybe they managed to bring memory usage down.
If you start Excel on your own pc, you will see that an empty sheet will take up to 5 Mb of memory (about the amount it uses within the session).
Could be that they are right, but normally if Microsoft says that you only need 128 mb (minimum) to let XP work, it wouldn't work either. Normally, if microsoft says something about memory usage, i always multiply it with 2. But indeed i could be wrong about the 64-128 Mb. Maybe your core software uses more memory. Problem with TS is (before going into a to long discussion ;)) is that you need to test it. Maybe 30 or 40 MB is enough for one session, maybe 10 Mb is enough (which i don't think) or maybe you do need 128 Mb.
TS 2003 is much better than TS 2000, maybe they managed to bring memory usage down.
If you start Excel on your own pc, you will see that an empty sheet will take up to 5 Mb of memory (about the amount it uses within the session).
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Normally i would say 128 Mb, because you actually are "recreating" a computer within a session.
Maybe this link helps..
http://www.microsoft.com/windows2000/server/evaluation/features/terminal.asp