The following is Tip #2 from "25 Exchange 2003 Tips in 25 minutes." This content is excerpted from Scott Schnoll's book, "Microsoft Exchange Server 2003 Distilled," brought to you by © 2004 Pearson Education, Inc. Publishing as Addison-Wesley Professional.
Requires Free Membership to View
Return to the main page for more tips on this topic.
Exchange 2000 included two tuning registry parameters specific to EVSs running in a cluster. Both parameters were designed to prevent a very busy SMTP resource in an EVS from starving other resources in the EVS such as IMAP4 and POP3. The tuning parameters were the SMTP % of threads and Additional threads per processor values, which were represented in the registry as follows.
Location: HKLM\System\CurrentControlSet\Services\SMTPSVC\Queuing
Value:MaxPercentPoolThreads
Location: HKLM\System\CurrentControlSet\Services\SMTPSVC\Queuing
Value:AdditionalPoolThreadsPerProc
MaxPercentPoolThreads was used to control the percentage of threads used by the SMTP
service, and AdditionalPoolThreadsPerProc enabled you to control the number of additional
threads that could be spawned on a per-processor basis. If you added either of these
registry entries to your Exchange 2000 cluster, you should remove them prior to upgrading
your EVSs to Exchange 2003.
Get more "25 Exchange 2003 Tips in 25 minutes." Return to the main page.
About the author: Scott Schnoll, an Expert on SearchExchange.com, is an MCT, MCSA and a long-time Microsoft Most Valuable Professional (MVP).
In addition to writing "Microsoft Exchange Server 2003 Distilled," he is a co-author of the upcoming "Exchange 2003 Resource Kit from Microsoft Press" and lead author for "Exchange 2000 Server: The Complete Reference."
Scott has written numerous articles for Exchange & Outlook Magazine, and is a regular speaker at Microsoft conferences, including MEC and TechEd, as well as industry conferences such as Comdex and MCP TechMentor, where he covers topics such as Exchange, clustering, Internet Information Services and security.

Join the conversationComment
Share
Comments
Results
Contribute to the conversation