Looking at metrics to understand use & requirments

  • Thread starter Thread starter Steve2
  • Start date Start date
S

Steve2

Is anyone using the performance metrics, either standard
or exstensible, to understand resource use of their
terminal server environment? Any guidence on what makes
sense to look at and what to avoid? A new farm of TS2K
is being built and expected to support 100s of users. No
idea if resources are enough. Thanks for any thoughts on
the matter.
 
How many users, what kind of hardware & what programs will they be using

Important items to watch are
Terminal Services - Active, Inactive & Total Session
Processor - %Processor Time (Individual & _Total
Memory - Paged Pool Bytes, Page Faults/Sec, Available MBytes, Comitted Byte
Paging File - %Usag
Physical Disk - Current Disk Queue Lengt
System - Processor Queue Lengt
Print Queue - Jobs Spooling (_Total

If the %Processor Time is constantly over 80% then you should either scale-out with another machine, add another CPU (if your server supports more than you're currently using) or use faster CPUs (if supported

If you're using MetaFrame you can split the number of users evenly among the servers in your farm by setting the Server User Load to a specific value, instead of the default of 100, i.e. if you have 4 Load-Balanced Citrix Servers and need to support 100 users you can set the Server User Load to report a full load when the number of server users reaches 34. Even if one node has to be taken offline you'll still be able to accomidate all 100 users, and one server won't get hammered while another is sitting idle

Hardware recommendations here
http://www.workthin.com/tshw.ht

How to articles here
http://www.workthin.com/tshta.ht

Patrick Rous
Microsoft MVP - Terminal Serve
http://www.workthin.co

----- Steve2 wrote: ----

Is anyone using the performance metrics, either standard
or exstensible, to understand resource use of their
terminal server environment? Any guidence on what makes
sense to look at and what to avoid? A new farm of TS2K
is being built and expected to support 100s of users. No
idea if resources are enough. Thanks for any thoughts on
the matter
 
Back
Top