| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: interpret v$mts
Really, the need for MTS is going away. It was designed for connection
pooling, but all of the application software vendors now provide connection
pooling with the app server. And it does it better than MTS ever did.
Given a choice, I always choose app-server connection pooling over MTS. It is just too hard to tune.
Tom Mercadante
Oracle Certified Professional
-----Original Message-----
From: MacGregor, Ian A. [mailto:ian_at_slac.stanford.edu]
Sent: Wednesday, June 02, 2004 10:03 AM
To: 'oracle-l_at_freelists.org'
Subject: RE: interpret v$mts
It would be helpful to know what is slow. Is it a query? Is it DML? Has the operation been traced. One cannot reduce all performance problems in an MTS environment to the need for more shared servers.
There is the possibility that the system is not suitable for MTS at all. MTS is meant for OLTP systems running properly tuned queries with small return sets, ideally one row. Other queries can tie-up shared servers for long periods of time, resulting in the need for more of them. If this is what's going on the choice to use MTS over dedicated servers needs to be re-examined. You can also look at having two connect strings, one for true OLTP operations and another with "server=dedicated" for your reporting queries.
MTS is to be avoided unless absolutely!! necessary. Has the number of simultaneous connetions to be supported been dtermined. Has the private memory needs for a shadow process been examined. Is the SGA properly tuned?
Ian MacGregor
Stanford Linear Accelerator Center
ian_at_slac.stanford.edu
-----Original Message-----
From: Jeroen van Sluisdam [mailto:jeroen.van.sluisdam_at_vrijuit.nl]
Sent: Monday, May 31, 2004 11:02 PM
To: 'oracle-l_at_freelists.org'
Subject: RE: interpret v$mts
Hi,
Just looking at these statistics is not enough Is the problem in the dispatchers or the servers? Try monitoring busy rate and available servers in this period. Anything known about the scope of this problem all users concerned?
Regards,
Jeroen
My set of sql to show mts info
set term off
set feedb off head on
col network heading 'Protocol' format A50 justify c
col busy_rate heading 'Busy Rate %' format 990.0 justify right
set term on
PROMPT Network statistics
Prompt
Prompt Busy rate should not be more then 50% select network,
sum(busy)*100 / (sum(busy) + sum(idle)) busy_rate
from v$dispatcher
group by network
/
set head off
PROMPT
Prompt
Prompt There should be servers and dispatchers waiting for new requests:
select 'Dispatchers ' ||status, count(STATUS)
from v$dispatcher group by status / select 'Shared servers:'||status,count(status)
from v$shared_server group by status / set head on
select * from v$mts;
exit
-----Oorspronkelijk bericht-----
Van: Barbara Baker [mailto:barbarabbaker_at_yahoo.com]
Verzonden: Friday, May 28, 2004 11:42 PM
Aan: oracle-l_at_freelists.org
Onderwerp: interpret v$mts
Dick (if you're out there) or someone who understands this mts stuff:
We're still having problems with slowness. I know you told me to just keep adding servers till the screaming stopped. I didn't listen too carefully. . .
I'm wondering if you can tell me if I'm interepreting this correctly. mts_servers is set to 30. At 10:09 this morning, my servers_highwater was 34. Here's select * from v$mts at 10:09
MAXIMUM_CONNECTIONS : 894 MAXIMUM_SESSIONS : 893 SERVERS_STARTED : 4 SERVERS_TERMINATED : 4 SERVERS_HIGHWATER : 34
-----------------
at 10:49, looked like this:
MAXIMUM_CONNECTIONS : 936 MAXIMUM_SESSIONS : 936 SERVERS_STARTED : 37 SERVERS_TERMINATED : 37 SERVERS_HIGHWATER : 52
-----------------
So does that mean that in 40 minutes, it started 33 servers? If so, isn't that, like, obscene??
I assume there's overhead associated with starting up these servers. Should I be setting mts_servers to something extremely high (like around 50)? It was within this 40 minutes that we saw extreme slowness. I don't know if I'm looking cause or effect, but it seems pretty strange to me that we needed an additional 30 servers in a 40 minute period.
Thanks for any advice on this one.
Barb
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html'unsubscribe' in the subject line.
-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request_at_freelists.org put
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html'unsubscribe' in the subject line.
-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request_at_freelists.org put
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.htmlput 'unsubscribe' in the subject line.
-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request_at_freelists.org
 |
 |