task: replace Oracle 11g with MySQL
current setup:
- Oracle 11g SE One
- about 80 tables
- stored procedures, PL/SQL are in place but out of scope for this question
- client applications do not use a generic access technique like JDBC but OCI libs
- applications have roughly 500 fixed SQL queries for Oracle
Management now wants to have outlined a migration draft. Don't bother about the schema or data migration.
I'm trying to figure out a rough estimate of time to spend for translating 500 queries. Naturally these queries range from very simple INSERTs/SELECTs to rather complex ones with subqueries, multiple JOINs and expressions like that:
TRUNC(EXTRACT(DAY FROM (afs.next - TO_TIMESTAMP_TZ('1970-01-01 00:00:00 UTC',
'YYYY-MM-DD HH24:MI:SS TZR'))) * 86400 + EXTRACT(HOUR FROM (afs.next -
TO_TIMESTAMP_TZ('1970-01-01 00:00:00 UTC', 'YYYY-MM-DD HH24:MI:SS TZR')))
* 3600 + EXTRACT(MINUTE FROM (afs.next - TO_TIMESTAMP_TZ('1970-01-01 00:00:00 UTC',
'YYYY-MM-DD HH24:MI:SS TZR'))) * 60 + EXTRACT(SECOND FROM (afs.next -
TO_TIMESTAMP_TZ('1970-01-01 00:00:00 UTC', 'YYYY-MM-DD HH24:MI:SS TZR')))) as \"next\"
Hint: sense and purpose of such expressions has to be read out of the existing queries because at creation time no one bothered about documenting details of the implementation.
I'm completely aware that there may be queries that cannot be translated as is because MySQL perhaps does not support some things.
Question: can you give a rough hint about the expected "throughput" you think an expert could be capable of? The expert will be a person skilled with MySQL, Oracle and the applications. What do you think how much time is required to "translate" a single average query and simply manually test it in the target database?
Please ignore the required testing of the whole application. Don't regard that we have to setup the target schema beforehand. Hints to and stories of failed migrations and disastrous problems are welcome.
Thanks!