"kart" <k***@discussions.microsoft.com> wrote in message
news:145489CD-74E6-46C7-9F1C-3B0462DAF8B0@microsoft.com...
> When we did stress testing, we found a strange behaviour in SQL Profiler.
> We
> validated the same after stress test is complete and found similar issues.
> Here is the detail.
>
> Our application is .Net based application, which call webservices (.NET).
> The webservice will can DAO components(as per architecture) to access DB
> thru' stored procedures.
>
> -> The event RPC:Completed is captured using SQL Profiler and it is found
> that the value as shown in Duration column (in Microseconds) is really
> huge,
> which is around 8 mts. But, the webservice completes in milli seconds.
> Even
> we evaluated the value using Milli Second in SQL Profiler(as SQL 2005 has
> milli to micro conversion in saving trace data) and the value is huge.
>
> Value as in microseconds(as in trace table) = 538912949 * 10^-6 = 538 sec
> =8.98 min.  But the start time and end time registered against this event
> is
>
> Start Time 2006-08-02 10:17:14.330
> End Time 2006-08-02 10:17:14.440 , which says the duration is only in
> millis
> sec.
>
> We faced similar issue in validating SQL profiler data, when we did
> profiling with single user.
>
> Could some one please help in identifying the cause for this.
>
> Note: We validated the final output of called procedure and it is as per
> expectation and proper.
>
> Regards, Kart
>

"Greg Linwood" wrote:

> Hi Kart
>
> What is your timeout setting?  If it's 8 minutes, this suggests that your
> webservice might not be closing a Recordset after the request has completed
> (connection pooling can cause this with older data access APIs such as DAO)
>
> The Duration column includes not only the execution time for a query, but
> also how long any server-side cursor is maintained open.
>
> Regards,
> Greg Linwood
> SQL Server MVP
>
> "kart" <k***@discussions.microsoft.com> wrote in message
> news:145489CD-74E6-46C7-9F1C-3B0462DAF8B0@microsoft.com...
> > When we did stress testing, we found a strange behaviour in SQL Profiler.
> > We
> > validated the same after stress test is complete and found similar issues.
> > Here is the detail.
> >
> > Our application is .Net based application, which call webservices (.NET).
> > The webservice will can DAO components(as per architecture) to access DB
> > thru' stored procedures.
> >
> > -> The event RPC:Completed is captured using SQL Profiler and it is found
> > that the value as shown in Duration column (in Microseconds) is really
> > huge,
> > which is around 8 mts. But, the webservice completes in milli seconds.
> > Even
> > we evaluated the value using Milli Second in SQL Profiler(as SQL 2005 has
> > milli to micro conversion in saving trace data) and the value is huge.
> >
> > Value as in microseconds(as in trace table) = 538912949 * 10^-6 = 538 sec
> > =8.98 min.  But the start time and end time registered against this event
> > is
> >
> > Start Time 2006-08-02 10:17:14.330
> > End Time 2006-08-02 10:17:14.440 , which says the duration is only in
> > millis
> > sec.
> >
> > We faced similar issue in validating SQL profiler data, when we did
> > profiling with single user.
> >
> > Could some one please help in identifying the cause for this.
> >
> > Note: We validated the final output of called procedure and it is as per
> > expectation and proper.
> >
> > Regards, Kart
> >
>
>
>

"kart" wrote:

> Hi Greg,
>
> Thanks for your reply.
>
> We are using ADO.NET. Please let me how to find the timeout setting that you
> are mentioning. In ADO.NET, the command timeout is specified as 30 sec.
>
> Eventhough, the duration is very high value, the Start  and End Time against
> this event is in milliseconds.
>
> Regards,
> Kart
> "Greg Linwood" wrote:
>
> > Hi Kart
> >
> > What is your timeout setting?  If it's 8 minutes, this suggests that your
> > webservice might not be closing a Recordset after the request has completed
> > (connection pooling can cause this with older data access APIs such as DAO)
> >
> > The Duration column includes not only the execution time for a query, but
> > also how long any server-side cursor is maintained open.
> >
> > Regards,
> > Greg Linwood
> > SQL Server MVP
> >
> > "kart" <k***@discussions.microsoft.com> wrote in message
> > news:145489CD-74E6-46C7-9F1C-3B0462DAF8B0@microsoft.com...
> > > When we did stress testing, we found a strange behaviour in SQL Profiler.
> > > We
> > > validated the same after stress test is complete and found similar issues.
> > > Here is the detail.
> > >
> > > Our application is .Net based application, which call webservices (.NET).
> > > The webservice will can DAO components(as per architecture) to access DB
> > > thru' stored procedures.
> > >
> > > -> The event RPC:Completed is captured using SQL Profiler and it is found
> > > that the value as shown in Duration column (in Microseconds) is really
> > > huge,
> > > which is around 8 mts. But, the webservice completes in milli seconds.
> > > Even
> > > we evaluated the value using Milli Second in SQL Profiler(as SQL 2005 has
> > > milli to micro conversion in saving trace data) and the value is huge.
> > >
> > > Value as in microseconds(as in trace table) = 538912949 * 10^-6 = 538 sec
> > > =8.98 min.  But the start time and end time registered against this event
> > > is
> > >
> > > Start Time 2006-08-02 10:17:14.330
> > > End Time 2006-08-02 10:17:14.440 , which says the duration is only in
> > > millis
> > > sec.
> > >
> > > We faced similar issue in validating SQL profiler data, when we did
> > > profiling with single user.
> > >
> > > Could some one please help in identifying the cause for this.
> > >
> > > Note: We validated the final output of called procedure and it is as per
> > > expectation and proper.
> > >
> > > Regards, Kart
> > >
> >
> >
> >