AND, (not but at all),
Don’t be hesitant to discard a few false starts. Make sure your timeline
includes time for a few fresh starts.
IF you have a way to simulate the activity including transactions, standard
reports, and most common ad hoc reports for a five month period of December,
January, February, March, and April for a January 1st year end, that will cover
a year end and a quarter end, which tend to be the most important. IF special
things happen at year and quarter ends and you don’t do this, you are in
extreme violation of #3 below.
IF you have time period comparison reports for, say, last year compared to
this year and the time to shove all data conflicts with the desire for multiple
trial conversions and dress rehearsals make sure you get enough months to make
the comparison reports realistic.
IF you have a lot of moribund data (data that is no longer allowed to be
changed) consider physically re-ordering, partitioning, and making such data
isolated in read only spaces as part of the migration. As needed for retrograde
plan or iso-plan performance outliers, extra specific indexes (and/or best
cluster factor and attribute ordering to match queries of existing indexes) may
be extremely useful to improve performance. The extra time to convert is often
time well spent in both size and speed.
IF you can simulate as suggested, that may tempt you to go light on suggestion
#2 below, ignoring the things that seem fine and just snapping the baselines
that have problems in your “test migration” when everything on the destination
is fresh and new. Don’t skip suggestion #2 unless you plan to have the old
machine available for years. Transactions can change the texture of the data
over time such that plan changes erupt differently in the new environment as
time passes. IF the old machine is gone, all those baselines are your time
machine to keep life BORING. BORING is good in IT operations. The excitement
should be innovative use of well running technology, not firefighting to keep
the technology from burning up.
Good luck. JL also has a blog thread of things that can go bump in the night
when you “didn’t change anything” that is a useful read to help THINKING
CLEARLY about things that erupt during your several dress rehearsals.
And remember: Two complete dress rehearsals with NO CHANGES before you pull the
trigger on the real migration. (Farnham’s Law of Sane Migrations, often
violated at great pain).
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On ;
Behalf Of Clay Jackson ("Clay.Jackson")
Sent: Saturday, November 13, 2021 4:36 PM
To: loknath.73@xxxxxxxxx; Oracle L
Subject: RE: Things to consider during upgrade/migration
What Mladen and Andy said – I would pay very serious attention to their wisdom;
born of many years 😊. ESPECIALLY the parts about
1. How different 19 is from 11 and the Exadata is from the HP
2. Collecting baselines for ALL SQL – the time you spend up from will save
more than that troubleshooting on the “back end”
3. TEST EVERYTHING
In my position, I talk to customers/prospects going through similar
upgrades/migrations and those who follow(ed) Steps 2 and 3 (and understood the
differences) had SIGNIFICANTLY better success than those who tried to
“shortcut” the process.
From: oracle-l-bounce@xxxxxxxxxxxxx <oracle-l-bounce@xxxxxxxxxxxxx> On Behalf
Of Lok P
Sent: Saturday, November 13, 2021 10:47 AM
To: Oracle L <oracle-l@xxxxxxxxxxxxx>
Subject: Things to consider during upgrade/migration
CAUTION: This email originated from outside of the organization. Do not follow
guidance, click links, or open attachments unless you recognize the sender and
know the content is safe.
Hello Listers, With respect to having a safe upgrade(say from 11.2 to 19C) or
migration(From HP to Exadata) experience with minimal performance issues. Is
there any guideline we should follow like setting up exadata system stats in
case the target database is going to be exadata, Or verifying dictionary
stats/table stats etc in a certain way. Want to know experts' views, if there
are any such guidelines?