Converting a 10.2–> 11.7

You'll be able to leverage the Pro2 version that uses CDC - it's a lot more efficient than the old version. I'd reach out to your Account Rep for details on documentation.

There is no 'may' about taking the system down. You will have to take the system down in order to upgrade the database. And if your database is still running Type II storage, now is a good time to consider a dump and reload.

Hi,

>>You'll be able to leverage the Pro2 version that uses CDC - it's a lot more efficient than the old version.

This is only if there is a license for OE CDC or Advanced Ent.

>>Where do do I find documentation  about Pro2?

The latest version of Pro2 and the documentation can be downloaded from your ESD account.

All documentation after Pro2 installation is in the directory <pro2dir>/Docs.

>>Any stuff I should consider when doing the convert?

It seems to me that the upgrade plan will depend on what kind of downtime is acceptable.

If there is enough downtime, then you can use the traditional method via conv1011 for DB + re-creation of AI replication + update Pro2 to the latest version. What Pro2 version do you have now?

If minimum downtime is required, then you can use Pro2 with Second Pass Replication. About this can be learned from the presentation of Mike Furgal on the PUG Challenge 410: Case Study: Platform and Data Migration with Little Downtime  where Pro2 used as the technology to do the dump/load for the platform migration.

Thanks, the customer has Pro2 Version 4.01 i think, and the system seems to be 50Gb in one db, 20Gb in another. Then they have a logdb that has about 100Gb, but that is not to be either replicated nor AI. It seems like the dump is taking 8Gb in 20min, I haven't tested the load yet, is to be continued :-)

I am not sure if the customer is licensed for OE CDC nor Adv Ent. I will have to check that.

note that the r-code version is different from 10.2 in 11.7. this means that you must recompile the entire application (might need to use keyword forget list). also, if you are going from 32-bit to 64-bit OpenEdge, then any shared libraries or dll’s you are using must be replaced with 64-bit versions.

the database conversion does not take long, but ai file format and replication are incompatible and must be done over.

one other thing: user names in _user can no longer contain a ”@“ character. but you can have “sql only” _user records.

-gus

Thanks, would it be signifisent better to do a dump&load compared to conv1011 ? They do not have any performance issues as it is now.

conv1011 is simpler and faster but not necessarily better.

if you make a new database via dump and load:

0) you get an opportunity to reorganize

1) you get a “clean” database with less dust, dirt, and grime inside it.

2) you will need more downtime. among other things, after loading, all indexes must be rebuilt.

if your data are not all in type ii data areas, then you should definitely dump and load.

whichever way you go, you will have to make a full backup before and after.