Oracle-DB: Cost of dedicated DB session connect/disconnect

 A generally assessed poor technique is to create an own dedicated database session for each atomic DB activity.

You may sometimes encounter such strategies like:

  • processing a large amount of items in a loop, each processing step in the loop creates a DB session, executes a small set of SQL statements and terminates the session
  • a polling process checks a SQL result one time a second, each in a new DB session
But what costs are generated by frequently connecting and disconnecting DB session?

The internal recording of database activity (AWR/ASH) has no answer because establishing the DB connection is not a SQL activity.

A simple comparison of system load gives a fuzzy hint to the price of connection creation.

Example:
  • An idle database instance on a single host with 4 older CPU cores (Intel Xeon E312xx, 2,6 GHz)
  • an external (not on DB host) SQLPlus client which executes a single "SELECT SYSTIMESTMP FROM DUAL" per DB session
  • Delay between the SQLPlus calls is calculated so that one connection per second is created and destroyed.
  • 6 Threads active each with one session creation per second
Result:
  • with idle database CPU load over 4 CPU nodes is in average 0.22%
  • with 6 threads creating and destroying sessions each second CPU load is 6.09%
  • additional io wait also occurs with 1.07% in average
  • so in average 5.87% of 4 CPU nodes are allocated by this 6 threads
  • Equivalent to 23.48% of one CPU node for 6 threads or 3,91% per thread
That means:
Connecting and disconnecting an dedicated Oracle DB session once per second costs approximately 4% of a CPU core of DB server.

This value in mind should help to consider if it's worth to change process behavior regarding DB session creation or not.
Solutions for that issue are to hold the DB connection over the lifetime of the application instance or use the various types of connection pools at client or server side.


Comments

Post a Comment

Popular posts from this blog

Oracle-DB: How to check for appropriate sequence caching

Image
A common pitfall while using sequences in Oracle-DB is forgetting to define sequence caching. The default for sequences is cache size = 0. That means, every sequence.nextval operation triggers a single write operation on dictionary table sys.SEQ$. Especially if frequently calling sequence.nextval uncached sequences may cause significant performance degradation. Frequently calling sequence.nextval in parallel sessions may lead to ramdom lock scenarios in library cache. Setting a cache size for a sequence may heavy increase performance of nextval-operations. There's a minimal memory overhead for caching a sequence because all that's needed is a counter to increase in SGA memory and a upper limit. If this limit is reached with the counter, the upper limit is increased by cache size and the new upper limit is written down once to SEQ$. The possible drawback for sequence caching is that you loose all your values between the current value and the highest cached value if yo
Read more

Common pitfalls using SQL*Net via Firewalls

It's very common that firewalls are terminating idle TCP sessions after a limited time, often after approximately one hour. Possible problem If you have a long running SQL or PL/SQL program and the firewall terminates your apparently idle TCP session this may lead to the following scenario: the database is not able to send result to client. After reaching the TCP-timeout the database server terminates the DB-Session the client program stays in socket read forever waiting for the database response Oracle's full solution: SQLNET.EXPIRE_TIME=x in sqlnet.ora Oracle's solution for this problem is setting parameter SQLNET.EXPIRE_TIME=x in sqlnet.ora where x are the minutes between keep alive packets on idle sessions that inform the firewall that this session is always alive. SQLNET.EXPIRE_TIME hast to be set in the sqlnet.ora of your RDBMS' ORACLE_HOME. Normally $ORACLE_HOME/network/admin. If you have a grid infrastructure setting SQLNET.EXPIRE_TIME in sqlnet
Read more

Oracle-DB: Identify excessive logon/logoff operations with short-running sessions

Image
Creating and destroying database sessions in loop may lead to significant database load. This blog post contains more about quantifying this load. Applications should use connection pooling at client or database side to prevent excessive session creation. But how can I identify the client processes causing this sessions? Scan DBA_Audit_Trail to identify context info If auditing logon/logoff-operations is activated for your database, scan DBA_Audit_Trail. You can do this by 2 was: Alternative 1: Use my tool "Panorama" to evaluate audit trail Panorama is available here. Go to menu "DBA general" / "Audit Trail": Select start and end time for selection Filter column "Action" for "LOGON" Set "Grouping" to "Minute" Execute selection via button "Show audit trail" Add columns of your choice to diagram by context menu (right mouse button) This way you can identify the machine, DB-user
Read more