Microsoft Exchange Server is a complex infrastructure to manage. Although Microsoft Exchange Server can be complex, there are some simple steps that can be taken to greatly improve system performance, keep downtime to a minimum and considerably reduce help desk complaints about sluggish e-mail response times. Let’s examine five simple steps you can take to tweak, configure and customize Microsoft Exchange Server for optimal performance in your organization.
Step No. 1: Offload spam and antivirus processing
Today, as much as 80 percent of e-mail messages received by an organization are considered spam. Many organizations run anti-spam and antivirus products in-house, in some cases using several layers of defense and solutions from multiple vendors. This approach not only increases the overall cost of managing the messaging system, but can lead to problems associated with deploying malware signatures or even basic operating system updates and configurations. This complexity can actually decrease security, not increase it.
The advantages of running antivirus and anti-spam services in the cloud are undeniable. Third-party service providers that specialize in messaging security can protect customers more cheaply and effectively than organizations can typically do it themselves.
These companies have invested large amounts of money, not only in research organizations and software, but also in the infrastructure to handle huge volumes of e-mail messages. In some cases, they even provide business continuity and disaster recovery services in the event of a catastrophe.
The biggest benefit of moving antivirus and anti-spam services to the cloud is the reduction of required bandwidth for SMTP traffic. Since e-mail is filtered by the service provider, the number of e-mail messages delivered to an organization’s network is dramatically reduced. This reduction often allows organizations to delay or cancel upgrades to network links and retire server hardware no longer needed to filter out spam.
Step No. 2: Move the search load from the server to the client
With the release of Exchange 2003 and Outlook 2003, Microsoft introduced “cached mode” for clients, which allows users to maintain a local copy of their data while being synced with the Exchange server. This is often confused with offline mode or PST mode where e-mail is simply uploaded and downloaded at preset intervals. Cached mode allows a user to be connected to the server in real time, while still using the local copy of their data to read/send e-mail and have access to accurate free/busy data.
Using cached mode eliminates the load on the server for reading and replying to e-mail messages since that work is done by users’ workstations. There are some downsides with deploying cached mode, namely around security. For example, data is residing on systems over which IT departments may not have full control. Consider using some sort of encryption technology to keep this data safe in the event that it is lost or stolen.
Optimize Maintenance and Backup Windows
Step No. 3: Optimize maintenance and backup windows
Exchange system maintenance is generally configured to run during off hours when there is reduced load on the server. The following is a simple checklist of four items that can be used to ensure that maintenance is running during an optimal timeframe:
1. For starters, review the configured maintenance window to verify that it is running during a period of low load and that it is not running during the backup window. Maintenance should never be run on a server while backups are taking place.
2. Implement a stop time that corresponds with the start of business when users add load to the server. The maintenance should be completed or halted before users need the server resources.
3. Check the event log and verify that maintenance is starting and completing successfully. Look for “Event ID 700” and verify that it is followed in sequence by “Event ID 701” and finally by an “Event ID 1221.” If “Event ID 701” does not appear, then online maintenance is not completing. Online maintenance should complete at least once per week.
4. Regularly monitor database white space. After online maintenance is complete, look for “Event ID 1221” which lists how much white space is available in the database. Many administrators will run offline defragmentation of the Exchange database when white space reaches 25 percent. This is not recommended. If a database has more than 50 percent white space, users should be moved to a newly created database. Only in very rare cases should the ESEUTIL tool ever be run.
How to Turbocharge Microsoft Exchange Server Performance
pagebreak title=Review Your Storage Configuration}
Step No. 4: Review your storage configuration
Exchange Server is a very storage-intensive application. In most cases, when an Exchange server is running slowly and the client experience is sluggish, this is due to undersized storage. There are three simple steps to find out if your Exchange server is suffering from undersized storage.
1. Find the I/O operations per second (IOPS) number. It is important to know how many IOPS users are generating against the Exchange databases in order to understand how much load is on the server.
To do this, use the performance monitor in Windows. Select the “Logical Disk/Disk Transfers per Second” counter. Take the average number of transfers during a busy period of four hours and divide this number by the number of users on that database logical unit number (LUN). This will equal the IOPS per user.
2. Find the capacity of your storage system. A typical 15K-RPM drive will produce approximately 160 IOPS per drive in a RAID 0+1 configuration. Simply multiply the number of spindles by 160 and that will equal the capacity of the storage system. For RAID four, divide the total IOPS capacity by four to account for RAID overhead.
3. Bring the numbers together. Take the IOPS-per-user number times the amount of users; if that number is larger than the available IOPS, then the storage system is undersized.
Step No. 5: Get rid of streaming backups
Streaming backup is supported by all versions of Exchange server and reads every page in the database and copies it to tape or to disk. This type of backup is effective for small deployments of Exchange (10GB or less). For Exchange deployments larger than 10GB, consider switching to a Volume Shadow Copy Service (VSS)-based “snap” backup technology.
Snap backups take a fraction of the time and have little to no impact on the user while a backup is in process. Also, snap backups allow for more frequent backup intervals. In most cases, an Exchange server can be backed up every four hours using snap instead of every 24 hours.
Lee Dumas is the Director of Architecture at Azaleos, and has been involved with Exchange and messaging for over 13 years-first as a core member of the Exchange development team at Microsoft and then in various consulting roles. He specializes in architecture and operations specific to Exchange and Active Directory, and is also proficient in other Exchange-related technologies such as UM, OCS, and mobile devices.
Most recently, before joining Azaleos, Lee spent seven years in “real world” deployment engagements in which he helped companies-ranging from a handful of seats up to multinational corporations with 60,000 or more seats-get Exchange messaging running smoothly. Lee is a pioneer MCA Ranger and has been part of the program since it first expanded outside of Microsoft four years ago. He can be reached at [email protected].