Thursday, May 6, 2010

What will Kill your Exchange Server so Quick ???

The performance troubleshooting can be tough time to time as it can be difficult to see or identify the bottleneck. The bottleneck and what to look for when it comes down to performance. Although seeing  Xmas balloon in outlook makes many of the Exchange administrator go bananas, the reality is , the heavy MAPI is easy to influenced and get worst with some many factors. This includes  client side troubleshooting as well as Server side troubleshooting etc.

Most of the issues I have seen so far caused by I/O related issues and problems in Exchange 2003 and 2007 as SAN is big part of complex Exchange and cluster configurations. The worst and most frustrating part is to deal with SAN  and its when problems balls down to IOPS.

Beside Exchange 2007 is complicated already, adding SAN into picture makes is even more complicated.

So imagine , most of the I/O will occur in the DISK resources and Exchange will love to hammer the hard disk logs and databases and outlook goes bananas.

If you are in the situation and need to find out what is going on here is your best friend and tool  ExTRA  ( Exchange Troubleshooting Assistant). You have to got to LOVE this tool as it is incredibly useful.


bare in mind the basic applies most of the troubleshooting cases. For most of the disk issues here what you need to know at the basic.

Check the performance counter below, they pretty much are explaining what will be the acceptable , upper limit. The bottom line is the PyhsicalDisk Averaged Disk sec/Write got to be <10ms Ideally , and the spikes should never be >50ms.

*** I have seen recently problems when Averaged Disk sec/Write >20ms as result , Outlook famous balloon pops up**** this isn't quite good experience and hard to explain client why it is happening (-: 


Also Check out “Perfwiz replacement for Exchange 2007” this is real nice resource, as pre defined counters are there for Exchange 2007

So what has changed in Exchange 2010? if I have to say one thing made me so happy , dealing with E210, I would say no more SAN and yes less complexity. Having only Exchange administrator eyes on the Exchange servers is great. We no longer need SAN expert to help us configuring or run and maintain Exchange 2010.

I think most of places who are using SAN do not know what to do if they get Exchange off the SAN, going for DAS and implementing DAG seems to be very optimal for most of these places. Huge saving are coming with using cheap SATA disk.

The  I/O  improvement in Exchange 2010 is 60 to 70 percent more compare to Exchange 2007, Investments done by exchange team in store schema and ESE (Extensible storage Engine ) have reduces IOPS by 70 percent more thus E210 is incredibly fast , bringing outstanding user experience not to mention tons of other improvements.

being able to implement JBOD ( bunch of disk ) yes we are talking about cheap SATA disk 7200RPM , with NO RAID , NO RAID controller is stunning. In recent when I see the saving in $$$$ I could not believe my eyes to be honest for large implementations. The saving $$$$ is HUGE. I will be honest I have not seen yet or done JBOD scenario but knowing from some of MVP colloquies  the JBOD is working perfect.

So the future of Exchange change a lot, the other competitors ( Gmail) were depressing Exchange in my personal opinion for the cost when MB cost is 4buks Exchange 2007 MB cost was bit over 16 dollars. Now I am asking myself what will be left when you take that away (-: , E210 MB cost can be less than 4 Bucks not to mention , so many incredible futures and incomparable messaging experience  with any other messaging application.

if you have not had your hands on E210, without wasting too much time go get evaluation copy and see with your own eyes , what does it look like the next generation messaging application.

Best regards,
Oz Casey , Dedeal
MCSE 2003, M+, S+, MCDST
Security+, Project +, Server +
Http:// (Blog
Http:// (Blog)
Http:// (Blog

No comments: