Well, I should have added that caveat I suppose. In our case, we were getting spikes of 4 seconds. On the plus side, Exchange wasn't corrupting the database, just freezing everything.
Spikes will still cause intermittent pop-ups unless you have clients in cached mode to mitigate this. Steven Peck On Feb 11, 2008 3:03 PM, Andy David <[EMAIL PROTECTED]> wrote: > Only if it's sustained above 50ms is it an issue. > Spikes are ok. > > > -----Original Message----- > From: Steven Peck [mailto:[EMAIL PROTECTED] > Sent: Monday, February 11, 2008 5:25 PM > To: MS-Exchange Admin Issues > > Subject: Re: Sluggish performance of Mail Store, Server sitting mostly idle > > Disk IO. Check your disk IO. If it is above 50 ms then you are toast > and you will have end users get the 'waiting on exchange server' > pop-ups. Exchange Best Practices analyzer can help you spot this. > > We have this issue currently using a Hitachi SAN. A major in house > project went live recently and it is also on the SAN. Performance on > Exchange went to hell. > > There are two points this can happen. > 1. You are sharing spindles on the SAN with other applications... (no > dedicated storage on a given LUN - common). The load on the other > application increases, you are doomed. > 2. You have a shared fabric to get to the SAN and your ports are over > subscribed or your Exchange server is sharing the same paths as your > other really busy IO intensive applications. > > We have both issues. We have dedicated storage now, but the path to > the dedicated storage is shared on the McData unit so no real help > yet. > > Steven Peck > > > > On Feb 11, 2008 11:23 AM, Robb Pickinpaugh > <[EMAIL PROTECTED]> wrote: > > Sean, > > > > I'll check the boot.ini switches hadn't thought of that yet. > > > > As for using Diskpart, yes I partitioned those disks following my initial > > setup SAs instructions very carefully. > > > > As for the dedicated spindles... we didn't have any way to determine the > > "damage" that was going to be done by the other app at the time. > > > > Now that I'm seeing the trend I'm running the ExBPA Health Check, with the > > 2 hour perfomance monitor, to see what comes back from that. > > > > I'll also be getting in touch with EMC support to find out how to get info > > out of the array itself as far as proving the overload. > > > > Then... and only then, can I try to get money for dedicated spindles. > > > > The admins of the other system have shut off the particular function that > > was causing the disk smash (for the moment). > > > > Still trying to make sure that Exchange in and of itself is configured > > correctly, and that there are no inherent performance problems besides the > > disks I have the store sitting on. > > > > Thanks for the additional info. > > > > ~ Ninja Email Security with Cloudmark Spam Engine Gets Image Spam ~ > > ~ http://www.sunbeltsoftware.com/Ninja ~ > > > > ~ Ninja Email Security with Cloudmark Spam Engine Gets Image Spam ~ > ~ http://www.sunbeltsoftware.com/Ninja ~ > > ~ Ninja Email Security with Cloudmark Spam Engine Gets Image Spam ~ > ~ http://www.sunbeltsoftware.com/Ninja ~ > ~ Ninja Email Security with Cloudmark Spam Engine Gets Image Spam ~ ~ http://www.sunbeltsoftware.com/Ninja ~