I would recommend looking at the following counters (straight out of the MEC 2002)
PhysicalDisk(drive:)\Avg. Disk sec/Read PhysicalDisk(drive:)\Avg. Disk sec/Write These two counters should be below 0.020 and if Write caching in array controller the sec/write should be less than 0.002 Anything between 0.020 - 0.050 seconds is a likely bottleneck Anything greater than 0.050 is definitely a bottleneck. We are currently utilizing IBM's FAStT Storage Solution as our SAN and are seeing Avg. Disk sec/Write's periodically as high as 10 seconds. Needless to say we have IBM out here trying to resolve it. I wish we had purchased the EMC....... -----Original Message----- From: Ed Crowley [mailto:[EMAIL PROTECTED]] Sent: Wednesday, February 05, 2003 11:03 AM To: Exchange Discussions Subject: RE: Exchange 2000 tuning That's where I would look. Ed Crowley MCSE+Internet MVP kcCC+I Tech Consultant hp Services Protecting the world from PSTs and Bricked Backups! -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of Ken Cornetet Sent: Wednesday, February 05, 2003 6:41 AM To: Exchange Discussions Subject: RE: Exchange 2000 tuning For logs I'm using an EMC LUN (is that still the right term for fibre-channel?) consisting of 5 mirrored pairs of 9GB disk slices. For the stores, I'm using two of the above LUNs in a Win2k stripe set. Each exchange server has one storage group consisting of two databases - both on the stripe set. The EMC admin tells me the remainder of the disks that I'm on hold low traffic stuff such as unix operating systems and low volume Oracle databases. Looking at the PhysicalDisk perfmon counters for the store disks, I'm confused. % disk time is averaging 257%. I assume that means the disk is not keeping up with requests, and some requests are sitting in queue? But % Idle shows an average of 65%. How can the disk be 257% busy and 65% Idle? Current queue length is averaging 3 (which sort of correlates to the 257% busy). Bytes per second is a paltry 380K. I've asked for tools to "look inside" the EMC to get stats on it's traffic, but we don't own anything yet (The EMC admin is in the process of buying something). -----Original Message----- From: Ed Crowley [mailto:[EMAIL PROTECTED]] Sent: Tuesday, February 04, 2003 6:35 PM To: Exchange Discussions Subject: RE: Exchange 2000 tuning How are disks configured? Ed Crowley MCSE+Internet MVP kcCC+I Tech Consultant hp Services Protecting the world from PSTs and Bricked Backups! -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of Ken Cornetet Sent: Tuesday, February 04, 2003 11:13 AM To: Exchange Discussions Subject: Exchange 2000 tuning Can anyone recommend any Exchange 2000 or Windows 2000 tuning parameters to consider tweaking? Or perhaps performance monitor counters to watch that would indicate problems? We recently completed migrating about 3500 users from Exchange 5.5 running on 4 quad 450MHz, 1GB machines to Exchange 2000 running on two dual 1.4GHz 4GB machines (win2k advance, /3GB switch). The old MSX 5.5 servers each had dual wide SCSI connections to an old EMC Symmetrix. The new servers have dual FC connections to a Symetrix. The old servers connected to the network via two teamed 100mbit ethernet lines. The new servers have one gigabit network link. Clients are running Outlook 98 and Outlook 2002. My mailbox was the first moved to one of the new servers. After it was moved, my outlook 2002 was lightening fast - faster than on the old servers (which wasn't bad at all). The problem is that now all the mailboxes are migrated, most everyone sees worse performance than on the old system. The odd part is that some people see no almost difference, while some see a significant drop in performance (as measured by the time it takes to display items in a folder). Viewing one's calendar seems particularly hard hit. I can't see any rhyme or reason to why some clients are impacted worse than others. It's not the client machine speed - I'm running an old Thinkpad 770Z - 366MHz PII and my performance is OK. It's not network - others around me (same network path to servers) see problems. CPU utilization on the servers very seldom goes above 20%, Pages/Sec typically sits at 0, but does bump to 10 or so. We were seeing lots of log stalls, but we raised the number of log buffers which did eliminate the log stalls, but didn't seem to improve performance any. We have 3 GCs (1.4GHz, 1GB) in the site where the E2K servers and most all of the users live. By all rights the two new machines SHOULD give even better total performance than the old four. The links to the EMC disks are faster (fibre channel), the EMC is newer (10K RPM disks, larger cache), the store is spread over more spindles, and gigabit ethernet. I know the new servers CAN provide more overall performance because backup time dropped by more than half over the old servers. _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED] _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED] _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED] _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED] _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED]
