Running Exchange 2007 SP2 using SCC connected to a SAN. We have most users on TierA storage, i.e. their LUN's are high quality, dedicated, fast disks. We have two more databases on TierC storage, shared LUN's with other apps. These are for temp users etc. We're looking at rolling out this model across our estates, since some business units don't want to pay for the high end storage if it's only part time users utilising Outlook. BUT - I notice that we are having latency issues with users on the TierC. Perfmon shows high Avg disk sec/read for the two DB's on TierC. This is to be expected. However, we also notice that sometimes when running Test-MAPIConnectivity, the latency results are high not just for the two DB's on TierC, but also some of the DBs on TierA. This only happened once we added the TierC storage. It only happens briefly, but does happen nonetheless. Is there *anyway* that poor RPC latency/disk queue on the TierC storage could affect users on TierA?

As I understand it, the terms Tier A, Tier B a d Tier C are completely arbitrary, relative only to each other, but providing no useful information to someone like me who's trying to answer your question. Basically what matters to Exchange is whether your storage subsystem can handle the IOPS thrown at it by Exchange, and that is generally determined by the number of physical disk spindles allocated to the databases or logs. Your description doesn't really tell us what's being stored on what servers. To answer your last question, the answer is "it depends". It depends on what load is placed on the storage, how the LUNs are carved out of drives, whether or not you share drives with other applications, the myriad inputs that determine server load, and lots more.Ed Crowley MVP "There are seldom good technological solutions to behavioral problems."

I apologize. TierA is meant to be hi quality, fast disks, TierC is lower quality, cheaper disks. TierA are dedicated disks for Exchange, TierC are shared with other apps. But the question was meant to mean this really - we are getting poor disk performance for the two DB's on TierC. Ok, that's fine, I understand how that can come about, because there probably aren't enough spindles and the ones that are there aren't good enough. But, since adding TierC, Test-MAPIConnectivity occasionallly times out for DB's on TierA. So the question is, is there any resource (apart from CPU/mem which are fine) that is shared, that could affect this? I would have thought issues related to RCP Latency on the TierC DB's/LUN's would be isolated?

How about the “Avg. Disk sec/Read” of the TierA disks for the problematic database? The article below will help Monitoring Mailbox ServersPlease remember to click Mark as Answer on the post that helps you, and to click Unmark as Answer if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.

Where are the HUB and CAS roles? Have you looked at them? John