Running a VM production file server for 150-200 concurrent users on ESX3.5? anyone doing it?

[LiamCurtis]

Hello all,

Wondering if anyone is doing it, has done it, or has experience or thoughts regarding running a vm file server for 150-200 concurrent users. We have an older compaq tasksmart and we are planning a consolidation and will probably purchase ESX3.5 as part of VI3. I would like to simply run a win2k3 vm file server and I'm wondering if the performance will be okay. We would most likely put our datastores on our Clariion CX500 (2gb fc) and run Proliant DL380G5 dual quad cores for ESX hosts.

thnks much in advance

[VMKR9]

Should run no problem at all, we ran a file server in a VM that was used for development, so was getting hit very hard, had over over 150 users connecting at once no problem and that was on a host running another 10 virtual desktops and servers...

[christianZ]

Should run w/o problems - have just installed a new nw 5.1file server for ca. 150 users as vm - runs better than the old ph.

[LiamCurtis]

Thanks much for the replies, all!

[eagleh]

VMKR9, may I ask your settings on this VM? vCPU and Memory? anything particular for such a heavy duty server? How busy the CPU works? Thank you.

[depping · Accepted solution]

Had about the same running at my previous employer. Had only 1 VCPU and 2048 of memory. Did disable updates of the last access time attribute for the NTFS filesystem which definitely made a huge difference.

Duncan

My virtualisation blog:

If you find this information useful, please award points for "correct" or "helpful".

[eagleh]

depping, thanks for the tip!

[ncasey]

I am running several file servers with at least 1000 concurrent users during the day with no problems.

VM's: only 1 VCPU and 2048MB memory

CPU pretty constant at 20% but I have access based enumeration enabled on the shares.

CISSP, GIAC, MCSE:Security, CCNA, CNA

[stormlight]

"disable updates of the last access time attribute for the NTFS filesystem which definitely made a huge difference"

How does this make a huge difference? Does it help with Disk IO? Proc or mem?

I am i correct that you will lose when the last time the file or fodler was edited? Are there any other downsides?

If you find this or any post helpful please award points

[eagleh]

My another concern for doing this is if my backup program would skip the files that should be backed up particularly in incremental mode. But there might be another "updates of the last modified" that is still enabled, so I need to find out which "updates" setting is actually affecting incremental backup.

[beckhamk]

I believe that backup programs use the last modified date. This is a 3rd item that is last accessed. Meaning and read/write will update the date time.

[eagleh]

just opened up a txt file, then closed it without any modification. the "Accessed" time didn't change accordingly. why? Having tried on different machines, the "Accessed" time is not changing at all. I am missing somthing here....

[beckhamk]

hmmmm maybe we need someone to explain what that truely means. I do see if you open and save a file that time is updated. It might be that the time is cached? or only certain circumstances is that updated. But after doing some more testing i think it does cache, or only updates the reads every so often. As I opened up explorer checked the last access time and then edit it without saving and then the time changed. But it doesnt seem to update 100% each time you read the file.

[eagleh]

Exactly. I noticed that too. I am lost. If it is cached, then it doens't make anything good. You are right, "save" back without modification will update "Accessed" time. "open it up then exit" won't. "Read" doesn't count as "Access"? That's funny.