When IBM migrated our DASD from 2105 to 2107 1st Q this year (on a
2064-106), FDRPAS is the tool they brought with them. CPU use by FDRPAS
was not an issue - device activity and channel activity could be,
especially if you have any specific devices that are already showing up
as a performance bottleneck (and having a faster processor actually
increases your exposure for DASD bottlenecks).
Do you depend on PAV support for performance on the 2105? One of the
limitations of FDRPAS 6 months ago was that it had to disable PAVs on
the source and target DASD volumes while it was migrating it from 2105
to the 2107. If you did the migration on a "hot" device at a time when
its workload required PAVs for acceptable performance, then applications
could see significant response degradation on the device for several
minutes, even if the DASD subsystems themselves had plenty of reserve
capacity. If FDRPAS still has this limitation and you depend on PAVs,
be forewarned and either chose slack periods for migrating critical
high-activity devices or monitor response closely during the migration.
We also did the migration without significant impact on the users.
Howard Rifkind wrote:
Hal,
I just went thru this 2 weeks ago.
Can you get your hands on FDRPAS? Worth every cent you pay for it.
We used it in prime time on an MP300/H-50, which is much slower then the box
you have.
I did it in prime time and it was a piece of cake, no one even know I was
doing it.
Hal Merritt <[EMAIL PROTECTED]> wrote:
We are on a countdown to move from a 2105-800 Shark to a 2107-921
DS8100. The motor is a z/890 2066-0A4. We are a 7x24 online shop with
some tight SLA's. All of our slack for years was consumed by a generator
failure.
We have four FICON paths to the Shark and two to the DS8100. The thought
is to swing one of the FICONs to the DS8100 making three each for the
transition.
The transition will be 'hot' using IBM services and the FDR product.
Question: are the channels faster in the DS8100? Should we stay with the
4-2 split until after the migration is complete?
Thoughts? Suggestions?
Thanks all.
...
--
Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
