Remco:

I appreciate all feedback, blunt or not.  I am relatively new to TSM but I
only work on windows client issues.  A separate team works on the TSM
storage servers and they are very experienced

The servers are loafing, they have 4 cores with 32 processors, and 384GB of
ram, not of which is anywhere near the limit.  The only bottleneck right
now is the 10GB interfaces in the exchange server and TSM storage servers
must pass through a 1GB embedded rack switch that I have been urging them
upgrade.  If we could get anywhere near 1GB network throughput on the
exchange backups that would be good.

I'm sure the storage servers are not under stress based on performance of
other backups we have running.

On Sun, Feb 18, 2018 at 1:03 PM, Remco Post <r.post AT plcs DOT nl> wrote:

> Hoi Tom,
>
> this might sound a bit blunt, but from what you’re asking I get the strong
> impression that this the first time you’re working with TSM. So I’m a bit
> anxious to give you any advise, fearing that it might lead to more problems.
>
> In general with performance issues I would look into the generic
> performance indicators of the exchange servers first. Secondly, check for
> any network bottlenecks between the exchange server and the TSM server.
> Thirdly you can look into the performance indicators of your TSM server.
> All with normal tools.
>
> > Op 17 feb. 2018, om 00:55 heeft Tom Alverson <tom.alverson AT GMAIL DOT COM>
> het volgende geschreven:
> >
> >>
> >>
> >> We are trying to speed up our Exchange backups that are currently only
> > using about 15% of the network bandwidth.  Our servers are running
> Windows
> > 2012R2 and Exchange 2013 CU15 with TSM 7.1.0.1 and TDPEXC 7.1.0.1.
> > Currently we are backing up 15 DAGS per Exchange server (we have multiple
> > exchange servers) and we are only backing up on servers that are standby
> > replicas.  Currently we are trying a 14 day schedule were we do a full
> > backup of a different DAG per day, and incrementals on the rest.  Even
> > doing this we are having trouble completing them in 24 hours (before the
> > next day's backup is supposed to start).
> >
> > I saw an old posting from Del saying to increase RESOURCEUTILIZATION on
> the
> > DSMAGENT.  Does that mean the DSM.OPT in the BACLIENT folder?  It was set
> > at 2.  Do either the buffers or buffrsize options make any difference?
> >
> > Also if we want to "parallelize" the backups does that mean separate
> > scheduler services for each one?  We currently use 14 different batch
> files
> > (for the 14 days of the cycle) with something like this:
> >
> > [day1.bat]
> >
> > tdpexcc.exe backup dag1 full
> > tdpexcc.exe backup dag2,dag3,dag4,dag5 incr
> > tdpexcc.exe backup dag6,dag7,dag8,dag9 incr
> > tdpexcc.exe backup dag10,dag11,dag12,dag13 incr
> > tcpexcc.exe backup dag14,dag15 incr
> > exit
>
> --
>
>  Met vriendelijke groeten/Kind Regards,
>
> Remco Post
> r.post AT plcs DOT nl
> +31 6 248 21 622
>