The document summarizes a system health session presented by Amit Banerjee. It introduces Amit as a senior engineer at Microsoft with expertise in databases. It then describes what system health sessions are, how they can help reduce analysis time and build visualizations from existing data. The rest of the document details what specific types of events and errors the system health session tracks to monitor instance health. It concludes by providing resources for learning more about system health sessions.

1. System Health Session
Amit Banerjee
Senior PFE
Microsoft Corporation

2. Amit Banerjee
• Senior Premier Field Engineer, Bangalore, India
• 6+ years with Microsoft
• Formerly with SQL Server CSS Escalation Services
About Me
• Technology expertise: Databases (SQL Server), Debugging
Highlights:
• Owner of www.TroubleshootingSQL.com
• Speaker at various SQL community events
• Co-author of Professional SQL Server 2012: Internals and
Troubleshooting
• Guest Author for SQL Server World Wide Users Group (SSWUG)
Contact Information
Blog: www.troubleshootingsql.com
Twitter: @banerjeeamit
Twitter: @TblshootingSQL
Email: troubleshootingsql@outlook.com
Facebook: www.facebook.com/TroubleshootingSQL

3. 1 6
POLL
Have heard about it
but never used it
Heard about
12 What is that?
12/22/12 3

4. THE NEED
Data that you can
use out of the
box
Reduce turn-
around time for Build data
post-mortem visualizations on
analysis of existing data
common issues
12/22/12 4

5. • The sql_text and session_id for any sessions that encounter an error that
has a severity >=20.
• The sql_text and session_id for any sessions that encounter a memory-
related error. The errors include 17803, 701, 802, 8645, 8651, 8657 and
8902.
What does it • A record of any non-yielding scheduler problems. (These appear in the
track SQL Server error log as error 17883.)
• Any deadlocks that are detected.
• The callstack, sql_text, and session_id for any sessions that have waited on
latches (or other interesting resources) for > 15 seconds.
• The callstack, sql_text, and session_id for any sessions that have waited on
locks for > 30 seconds.
• The callstack, sql_text, and session_id for any sessions that have waited for
a long time for preemptive waits. The duration varies by wait type. A
preemptive wait is where SQL Server is waiting for external API calls.
• The callstack and session_id for CLR allocation and virtual allocation
failures.
• The ring_buffer events for the memory broker, scheduler monitor,
memory node OOM, security, and connectivity.
• System component results from sp_server_diagnostics.
• Instance health collected by
scheduler_monitor_system_health_ring_buffer_recorded.
• CLR Allocation failures.
• Connectivity errors using connectivity_ring_buffer_recorded.
• Security errors using security_error_ring_buffer_recorded.
12/22/12 5

6. System Health Session
Resources
http://msdn.microsoft.com/en-us/library/ff877955.aspx
System Health Session Posts
http://troubleshootingsql.com/2012/09/24/system-health-session-
and-beyond/
System Health Session Custom Reports
http://troubleshootingsql.com/2011/09/28/system-health-session-
part-4/
12/22/12 6

7. www.troubleshootingsql.com
@banerjeeamit
@TblshootingSQL
troubleshootingsql@outlook.com
Facebook.com/TroubleshootingSQL
© 2012 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other
countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must
respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided
after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION