Ansicht
Dokumentation
10736 - DB2 - Two tier vs 3 tier
SUBST_MERGE_LIST - merge external lists to one complete list with #if... logic for R3up BAL_S_LOG - Application Log: Log header dataThis documentation is copyright by SAP AG.
DB2 - Two tier vs 3 tier
Hi Craig,An app server is either calling the db directly (2-tier) or it is
using a server job (3-tier). It is possible to run in a 3-tier
mode even within the same LPAR if you have a problem with your TCP
config. The Local host name (from CFGTCP opt 12) is what SAP uses
to determine the App server name. The DB host name is set in your
instance/default profile. If there is an uppercase/lowercase
mismatch, the code thinks they are different machines and will use
the network and server jobs to connect from the App server to the DB
server, even though they are really on the same LPAR. Hopefully
that's the case here, because it is really easy to correct.
One way to check is to look in one of the developer traces. The one
for the first work process will do. It's found
in /usr/sap/<sid>/....<inst#>/work/dev_w0. What SAP is using for
the local and db host names can be found in that file.
If it isn't that, I'd check normal instance tuning knobs again.
Ron
Opinions here are my own, not IBM's
--- In DoNotReply@consolut.eu, "sapbasisguy"
<craig_welchZb...> wrote:
> Hi Ron,
>
> Do you have any insight into the 2 1/2 tier architecture? I
noticed
> when running an application server instance on the same LPAR as
the
> central instance, the performance was degraded in the application
> server instance. The application server instance I built only had
> dialog, spool and batch in it. Would not having update work
processes
> defined also in the application server account for the degredation?
>
> Thanks,
> Craig
> Opinions my own.
>
> --- In DoNotReply@consolut.eu, "rschmerbauch"
<rschmerbZu...>
> wrote:
> > Yes John, you are correct.
> > On iSeries 2-tier we have an advantage as the SAP jobs can call
> > direct to DB apis, without the need for shadow(server) jobs.
On
> > iSeries 3-tier, there is a network time component, and the shadow
> > (server) job component...and this usually results in a 1.5x to
2x
> > response time increase over iSeries 2-tier.
> >
> > On other platforms without integrated DBs, the DB is just
another
> > application and it has it's own server jobs. So both 2-tier
and 3-
> > tier have a server component...so the delta between 2 and 3 tier
is
> > less noticable.
> >
> > Sometimes we get criticism from time to time from consultants
> > about "slow" 3-tier performance as compared to 2-tier, but it's
> > often just because we don't have the silly server job slowing us
> > down on 2-tier like other platforms.
> >
> > --- In DoNotReply@consolut.eu, "ibmjw014" <jbwassocZh...>
> > wrote:
> > > Hello,
> > >
> > > In our testing of IO bound programs we see the ABAPs running
> > > about twice as fast on a 2 tiered test system than on our 3
> tiered
> > > production system.
> > > Somewhere I remember reading that on a two tiered system DB2
> > > unlike oracle, bypasses the database shadow processes making
the
> > DB2
> > > database access extremely fast.
> > > I have not been able to find any documentation from either
IBM
> > or
> > > SAP to confirm this. Any help or comments would be gratly
> > appreciated.
> > >
> > > regards
> > > John
Durban Tours - Südafrika Safari
SUBST_MERGE_LIST - merge external lists to one complete list with #if... logic for R3up General Material Data
This documentation is copyright by SAP AG.
Length: 4467 Date: 20240520 Time: 191001 sap01-206 ( 3 ms )
