SA16 - Alternative to DBPing (with JDBC) / SQL connection performance

[Former Member]

Hey Guys,

DBPing seems to be a nice option to get the SQL Connection-Performance (Client to Server) In the Sybase-Documentation http://dcx.sybase.com/index.html#sa160/en/dbadmin/dbping-ping-dbutilities.html they write thats not possible with jdbc / jConnect (TDS) Connections.

Anyone know another good way to measure / to get the connection Performance from client to Server. I am often on the lookout for the bottleneck (network or database).

I think about different measuring points

• Client -> Server Server -> Database
• Database -> Query First Fetch
• Database -> query result set Display
• the result set to the client

Update: A specific case, I had already been described in the past.

http://sqlanywhere-forum.sap.com/questions/22413/checkpoint_time-option-multi-programming-level-slow...

In the above case, certain queries took in a high availability environment up to 30 seconds for no apparent reason.

We have now found the cause. It was a badly configured not upgraded Switch. Due to the high number of packet losses, the switch repeatedly had to request missing packets to request fully forward.

After the update, query times were all back in order.

In December 2014, the customer has also updated to Sybase16 and now has no more problems.

How can I find something like this in the future? The database has shown me slow queries, but not because it was the database but the network.

Is there a tool or a program with which I can figure this out. The tool could record all the way and then split into different sections such as the network of time, database and network way back

[Former Member]

..if that makes any sense... (the list with the different measuring points )

[graeme_perrow]

@pmiller - I took the information from your answer and added it to the original question.

[VolkerBarth]

Hm, my idea had been to add it as answer to the cited question (i.e. http://sqlanywhere-forum.sap.com/questions/22413)...

[Breck_Carter]

It may not make any sense to care how dbping connects, if you are mainly interested in measuring the performance of your environment. If you are debugging problems with connecting, then certainly you care about embedded SQL versus ODBC versus TDS (and would be within your rights to ask for dbping to support TDS), but the performance numbers reported by dbping will never match match your own client application... so you can't compare dbping numbers with your application numbers.

What you can compare, is the DIFFERENCE in dbping numbers when you make changes to your database and to your network, and (presumably) the difference will be the same regardless of the connection method. For example, if you make a network change, and dbping reports 100% better throughput using embedded SQL, it makes sense for you to experience (something like) 100% better throughput with your application even though it connects via TDS. If your application doesn't experience the same improvement, it probably won't be because of embedded SQL versus TDS.

Please let us know what you are looking for, from a business point of view... perhaps I have misunderstood your needs.

Performance Statistic           Number            Total Time    Average
----------------------------    --------------    ----------    ------------
DBLib connect and disconnect      200 times       3200 msec        16 msec
Round trip simple request        8950 requests   10112 msec         1 msec
Send throughput                 71200 KB         10112 msec      7041 KB/sec
Receive throughput              62700 KB         10112 msec      6200 KB/sec