Microsoft Exchange Server 2003 - Performance Monitoring
Introduction to Exchange 2003 - Performance Monitoring Advice
Network performance monitoring is a complex subject, some would say a black art. Many network administrators, who are highly skilled at other Exchange 2003 server tasks, have relatively little experience with Performance Monitor logs. By finding and removing bottlenecks, not only will you speed up your server, but you will also learn a great deal about how Exchange processes and objects operate.
Topics for Performance Tuning in Exchange 2003 Server
Set Yourself Goals - make a plan
So begin your performance monitoring project by writing down your goals.
Collect background information
Exchange 2003 relies on Windows 2003 for controlling memory, CPU, disk and network. Find out the specifications for your Exchange Servers, some people are shocked at how little RAM they really have; others are pleasantly surprised that they have 4 processors not 2.
Exchange 2003's store.exe usually consumes most of the server's available memory. So, always check for bottlenecks in this order, memory, processor, disk and network. Seek the root cause of any bottleneck, for example high disk activity could mask an underlying memory shortage. What is happening here is the disk is paging, rather than performing useful work such as writing to the exchange databases. Understand that a spike of activity is normal, whereas continuous high levels indicate a bottleneck.
Look for solutions
SolarWinds' Network Performance Monitor will help you discover what's happening on your network. This utility will also guide you through troubleshooting; the dashboard will indicate whether the root cause is a broken link, faulty equipment or resource overload.
What I like best is the way NPM suggests solutions to network problems. Its also has the ability to monitor the health of individual VMware virtual machines. If you are interested in troubleshooting, and creating network maps, then I recommend that you try NPM now.
Interpreting Log Data
The biggest problem that you face is interpreting the log data. What everyone wants is more 'rules' such as : If Processor % time is greater than 80% = bottleneck. Unfortunately, interpreting the log charts produced is rarely that simple. It's relatively rare that you can pin down a threshold with one value and say: 'more than x and you have a bottleneck'.
In my case, it's not reluctance to tell you what I know; more that Exchange is so complex that it's difficult to give hard and fast advice without knowing your situation. For example, take my rule of thumb for a queue, any queue; Guy says more than 2 object in a queue means you have found a bottleneck. Now take the case of where the disk queue has a value of 3. I can hear you thinking: 'Guy says we have a bottleneck', however it turned out the server had quad disks with 4 spindles, so the queue value should really be 0.75 - well under the threshold. This is why you really must examine many factors before you can be sure you have isolated the bottleneck.
Is performance monitoring a science or a black art?
One of the pleasant side effect of analyzing Performance logs, is that the counters teach you so much about how the Exchange email system work. For example caching, memory, information store.
When you begin climbing the Performance Monitor mountain, the first few steps are really steep. Don't give up because as you build skill, so the path gets easier.
Most people find Performance monitor is a difficult tool to master. If you persevere with creating logs and analysing the data, you will be rewarded with a more efficient Exchange 2003 server. Moreover, as you experiment with Performance counters, so you learn about the underlying Exchange process, for example information store, memory and routing.