Gerald,

A few points and comments:

- Keep in mind that the Domino transaction log is one large
  logical log file (in 64M segments) that contains entries
  for each and every logged database. It was built for
  not only built for crash recoverability... but also to
  enhance performance during standard production operations.

- Data Protection for Domino actually uses a log "prefetch"
  algorithm. That is, while the Domino server is replaying the
  current transaction log, Data Protection for Domino uses
  another set of threads to start restoring the next logical
  transaction log in sequence. The log prefetch and the log replay
  are performed simultaneously to maximize efficiency.
  Data Protection for Domino prefetches one log ahead.
  If the Domino server starts asking for logs file out of sequence,
  prefetching is automatically turned off.

- You can also turn "prefetching" off to see if prefetching is actually
  detrimental in your environment. You can turn it off by adding
  the following to "domdsm.cfg" (or whatever configuration file
  you are using):
         LOGPREfetch NO

- It is true that transaction log replay can be quite lengthy.
  Many customers set up a procedure to perform full database backups
  more often so that the log replay time for restores is smaller
  because there are less transaction logs to replay.

Thanks,

Del

----------------------------------------------------

"ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU> wrote on 10/28/2004
07:58:50 PM:

> A question regarding TDP if anyone knows for sure..
>
> The way the Notes TDP replays incremental logs is it pulls down a
> single 64MB TXN file at a time, replays those transactions, then pulls
> the next one down (restores). It's a pretty slow process that spends
> most of the restore time actually replaying the transactions. The
> restore of those single TXN files is quick and dirty. If a restore
> takes 12 hours to do the incremental portion, most of those 20 hours
> isn't restoring the actual TXN files but rather replaying the data
> between each TXN file.
>
> Does anyone know if theres a quicker way? Perhaps i can selectively
> restore all TXN files and dump them to disk someplace, then initiate a
> Notes command to replay them straight off disk, taking TSM out of the
> equation and more importantly, freeing up the drive for another job?