Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - Andrew H

Pages: [1]
1
Documentation / TSM Connection Pooling Introduced with 8.5.0.6
« on: March 26, 2013, 03:59:34 PM »
As of 3/26/2013, this is an undocumented feature.  Be aware the feature is turned on by default.  Via a PMR, IBM provided the following response:

CMOD connection pooling logic

CMOD recently implemented a form of connection pooling for TSM.  It is available in CMOD versions 8.5.0.6+ and 9.0.0.0+.
Up to 100 concurrent sessions total per CMOD library or object server
Short-lived connections (15 idle seconds)
Once the maximum CMOD connections are in use, new connections are added by closing the oldest current connection when it goes idle and adding a new connection to the pool

Prior to implementing connection pooling CMOD would perform one function per connection.  For example, to retrieve an object, CMOD would open a connection to TSM, query for the object, retrieve it if found, and close the connection.  For a CMOD system which is not under load, this is still the behavior.   

However, for systems under load, if TSM requests are coming in faster than 1 every 15 seconds, then the connection pooling will start to be noticed in that the connection to TSM will no longer be closed after the function is completed and be reused if idle.  Otherwise multiple simultaneous connections will be created.

The reason these connections are short-lived is to prevent media contention on the TSM server.  We noticed during testing that keeping the connections open permanently caused volumes to be locked until the volume timeout.  Since the minimum volume idle dismount time is 1 minute, this proved unacceptable in load testing.   

2
Documentation / Server Based Searching?
« on: October 29, 2012, 06:01:58 PM »
Hey all
Andrew H here from PFG.  In a few of the IOD sessions regarding 9.0's new Text Search capabilities, the old method of "Server Based Searching" was discussed. I am unfamiliar with this option and in the past, have simply run ARSDOC to retrieve documents and then a search utility against the content.  Can anyone comment on how to use "Server Based Searching"  Just today, I received a request to pull back a significant amount of content so thought I'd try this "new" old method...

Andrew

Pages: [1]