Maximum ‘anything’ when running a high number of processes on Windows

capacityperformancewindows-server-2012-r2

I have a series of well optimised services written in house using .Net running on many 512GB RAM server (a mix of Xeons including E5-2697 V2 etc). Is there a maximum handle, process, thread count that would contribute to services becoming unable to stop and start? Haven't got anywhere with Microsoft support to date.
Currently the machine is using approx :

81GB RAM
1,000 processes (about 400 user accounts for isolation between each service and child processes)
14,500 threads
1,000,000 handles

We also seem to max out around 550 users on RDP machines on Windows 2012 R2.
Is this known/expected?

Machines have plenty of RAM and plenty of CPU.
We have even tried running virtual machines, but splitting across machines seems to still hit this 550 user limit.

Best Answer

Yes - Mark Russinovich wrote a whole series of posts, unfortunately back in 2008 so some of it might be out of date. Link-and-run answers aren't great, but he wrote a lot on the topic that I cannot easily summarize here. He also shows a bunch of tests that you could run to see if some of the limits in 2012R2 have changed.

http://blogs.technet.com/b/markrussinovich/archive/2008/07/21/3092070.aspx

I have never seen anyone get over 500 interactive RDP users onto a single Windows machine. Hell, I don't think I've ever seen more than 80. Is RDP really the best choice for interaction with your services?

Related Solutions

Linux Server Performance – What Limits Maximum Number of Connections?

I finally found the setting that was really limiting the number of connections: net.ipv4.netfilter.ip_conntrack_max. This was set to 11,776 and whatever I set it to is the number of requests I can serve in my test before having to wait tcp_fin_timeout seconds for more connections to become available. The conntrack table is what the kernel uses to track the state of connections so once it's full, the kernel starts dropping packets and printing this in the log:

Jun  2 20:39:14 XXXX-XXX kernel: ip_conntrack: table full, dropping packet.

The next step was getting the kernel to recycle all those connections in the TIME_WAIT state rather than dropping packets. I could get that to happen either by turning on tcp_tw_recycle or increasing ip_conntrack_max to be larger than the number of local ports made available for connections by ip_local_port_range. I guess once the kernel is out of local ports it starts recycling connections. This uses more memory tracking connections but it seems like the better solution than turning on tcp_tw_recycle since the docs imply that that is dangerous.

With this configuration I can run ab all day and never run out of connections:

net.ipv4.netfilter.ip_conntrack_max = 32768
net.ipv4.tcp_tw_recycle = 0
net.ipv4.tcp_tw_reuse = 0
net.ipv4.tcp_orphan_retries = 1
net.ipv4.tcp_fin_timeout = 25
net.ipv4.tcp_max_orphans = 8192
net.ipv4.ip_local_port_range = 32768    61000

The tcp_max_orphans setting didn't have any effect on my tests and I don't know why. I would think it would close the connections in TIME_WAIT state once there were 8192 of them but it doesn't do that for me.

Ubuntu – Why is Apache running so many processes? Excessive RAM here

Apache is running multiple processes to have them ready when a client request comes in. Spawning a server process is slow, so it's best to have one waiting for a client.

For memory usage, you should take into the account RES size (as displayed by top), which is the amount of physical memory used by the task.

Why do you think you have too many apache2 processes using too much memory? What do you expect? Why do you expect so?

Best Answer

Related Solutions

Linux Server Performance – What Limits Maximum Number of Connections?

Ubuntu – Why is Apache running so many processes? Excessive RAM here

Related Topic