MCS UK Unified Communications Blog

All things UC related from Microsoft Consulting Services UK : architecture, best practice, news for Exchange and Lync, both on-premise and cloud

Microsoft Exchange Server Jetstress 2010 Beta

Microsoft Exchange Server Jetstress 2010 Beta

  • Comments 1
  • Likes

I have just received a note from Erin Bookey announcing the public beta of JetStress for Exchange 2010.  You can download your copy here…

http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=13267027-8120-48ed-931b-29eb0aa52aa6#tm

This is the announcement…

Jetstress 2010 (Beta Preview) – Available Publically!

9/01/2009

Questions:  jetfeed@microsoft.com

Download location:  Microsoft Exchange Server Jetstress 2010 Beta (64 bit)

Jetstress has been updated to support the new IO characteristics of Exchange 2010. This build is now available publically.  Please read this email carefully.

Feedback

Please send feedback (including bugs) to: jetfeed@microsoft.com.   For bug reporting, please include the following:

  • Your configuration xml file
  • Your Jetstress log file
  • Any pertinent html reports

CHM Help File

The Jetstress 2010 Help chm has not yet been updated.  The chm content is still Jetstress 2007 and does not reflect any change made to Jetstress 2010.  See “New Functionality” below for details on new Jetstress 2010 functionality.  Send email to jetstres@microsoft.com with any usage questions.

Scenarios

Jetstress dev/test has prioritized validating Jetstress 2010 with E2010 versions of ESE vs. testing legacy ESE versions (E2K3 and E2K7).  The goal of Jetstress 2010 is to completely replace Jetstress 2007 so it does support E2K3/E2K7 ESE binaries, but those scenarios are much less tested at this time.  Please focus Jetstress 2010 with E2010 ESE binaries and use Jetstress 2007 for E2K3/E2K7 testing.  Jetstress 2010 requires the most recent version of E2010 ESE binaries.  Jetstress 2010 supports Windows 2003 and higher but E2010 supports W2K8 SP2 or W2K8 R2 (W7).  Please use W2K8 SP2 or W2K8 R2 (W7) with Jetstress 2010 when using E2010 ESE binaries.  Only the 64bit version of Jetstress 2010 has been provided.  32bit Jetstress 2010 will be available in the final release.

New Functionality

  • Here is a list of the most significant changes to Jetstress 2010:
  • Updated with Exchange 2010 Mailbox IO profile (32KB page size, avg DB read size, avg DB write size, avg Log write size etc..)  This profile is not yet final and is subject to change between now and E2010 release.
  • Database duplication is now multi-cast.  This dramatically reduces the time to prepare databases for Jetstress testing.
  • Moved away from using Logical Disk performance counters to using MSExchange Database IO counters for IO measurement.  This now allows placing databases and logs on the same volume.  This also solves the usability issue of Jetstress not working correctly when there is a Logical Disk performance counter problem.
  • Log check summing is now multi-threaded (logs from different ESE instances check summed in parallel).
  • Maximum number of Databases per test increased to 100 for E2010 ESE binaries.
  • Log replication IO is now simulated based on E2010 HA architecture.  You can simulate a maximum number of 16 DB copies to drive the log replication IO (Log read IO) via xml configuration or UI in Jetstresswin.exe.  The IO generated by simulating log shipping as well as the configuration is provided in the HTML report.
  • Background Database Maintenance (Checksum) is now simulated for E2010 test configurations. This is the default behavior (like E2010) and the work BDM does during the test is reported separately from the normal transactional IO in the HTML report.  This can be disabled via the xml configuration or via a checkbox in Jetstresswin.exe.
  • Traces now contain date and time information for every entry;
  • More informative messages and warnings are provided prevent admins from using mismatched ese/eseperf and have further problems with performance counters errors when generating test reports;
  • Only the minimum necessary number of databases is created/replicated (as opposed to 5 databases per ESE instances);
  • ESE performance counters are now unregistered/registered back if it is detected that the current registered version is inconsistent with the version Jetstress will use. Restarting Jetstress is still required;
  • JetstressCmd and JetstressWin are now consistent in overriding the test type from Performance to Stress if the duration is longer than 6 hours.
  • System and Application events are now saved off per Jetstress run for time of run (along with html reports and perfmon log).

JetstressWin UI

  • “Select Test Type” page now displays a “Run database maintenance” checkbox. Only visible for E2010 and not visible is the test type is “Streaming Backup”
  • “Define Storage Group” page was renamed to “Define Database Configuration” and the Storage Group concept was removed.  To use Jetstress 2010 with E2K3/E2K7 binaries where multiple databases are desired per SG, the admin was now set the Jetstress 2010 log paths for DB instances they want to be in the same SG to the same path.  E2010 only supports only one DB per log stream so this doesn’t apply.
  • A new “Number of database copies” checkbox was added to the “Define Database Configuration” page. Only supported for E2010.

JetstressCmd UI

  • Added “/DBCopies <# of copies>” to allow the admin to specify the number of database copies to simulate E2010 HA log replication IO. Only E2010.
  • Added “/DoNotRunDBM” to allow the admin to not run database maintenance (default is to run). Only E14+;

Test Reports

  • HTML reports now break out IO in to the following table categories:
  • Transactional IO Performance:  This table just reports the transactional IO performance (e.g. Outlook generated IO).  This report is generated based on subtracting the Background DB Maintenance IO and Log Replication IO from the Total IO.  This report provides that actual DB IOPS generated and DB/Log latency that is required to judge a Jetstress performance test as Pass/Fail.
  • Background Database Maintenance IO Performance:  This table reports just the IO generated due to background maintenance.
  • Log Replication IO Performance:  This table reports just the IO generated from the simulated log replication.
  • Total IO Performance:  This table reports the total IO generated during the Jetstress test.
  • Added read and write average sizes for databases and logs.
  • Database configuration is now shown in the test report.

Thx,

The Jetstress Development Team

Posted by Neil Johnson, MCS UK, MCM Exchange 2007

Comments
Your comment has been posted.   Close
Thank you, your comment requires moderation so it may take a while to appear.   Close
Leave a Comment