Progress with DoubleTake

idingsdale

New Member
Hi All,

I'm pretty new to Progress, so apologies if any of this seems obvious.

We have a couple of databases on Progress 10.1A and want to replicate them to an off site server over a WAN link. Progress Replication is too expensive and as part of our wider DR strategy we wil be purchasing DoubleTake software (this is Windows 2003..).

What would be the most sensible strategy for ensuring the database is recoverable and consistent? Replicate AI, BI and DB files? Replicate AI files only and apply them to the offsite database according to a schedule? Replicate the DB files only?

Thanks for any advice!
 
I don't know anything about "DoubleTake" but it is fairly straight-forward to use after-image files to keep a replicated database in a "warm spare" status.

You still need to license it though. Progress charges 50% of the production license cost for a "dr" database license.
 
Hi, Thanks for your reply.

Doubletake is replication software, it replicates files as they are changed at the block level. DoubleTake have said they can support Progress, but not if the DB is in "Direct IO mode" as the DoubleTake service does not see the writes to disk.

As the replicated data would be to an offline image, not a running server, I don't believe we should require any additional licensing.

If we did go down the DoubleTake route, would a DB file without the AI or BI files be recoverable?

Are there any guides to replicating using the AI files? Presumably it needs to be configured with relatively small extents which are then copied and applied to the warm spare db?

Thanks again
 
I looked up DoubleTake online. Their web site isn't terribly informative regarding how the software works so I can't be sure about the details.

You can use tools like EMC's SRDF to do this sort of remote replication safely. DoubleTake seems similar in concept but operates at a higher level than that so it may, or may not work. For such tools to work properly they have to be very careful about preserving the order of changes. They not only have to capture byte by byte changes but they need to do it across all of the parts of the database synchronously. All of the data extents, all of the bi extents and all of the ai extents.

It might work. But it would take a lot of testing to make me confident.

Using after-imaging is, as I said, fairly easy. The simplest approach is to use variable length ai extents and to switch extents based on time. So every 15 minutes (or whatever period you're happy with) switch extents, copy the full extent to the warm spare and roll it forward.
 
Hi, I have been involved with over 1600 Double-Take installations and several have involved progress databases. Double-Take uses write order preservation to keep order of any database transactions. With typical database protection you normally have to replicate the database and the associated log files. I know there is an application notes for many db's but I don't see anything specific in the Double-Take knowledge base on progressive and there are no issues that I am aware. You can always call Double-Take support if you have further questions and they may be able to provide you some more specifics. I know Double-Take also has a usergroup of other double-take customers who may have addional information to share.

Take Care,
 
Thanks for the response, its reassuring to know I'm not the first to do this!

Do you know where I might find the DoubleTake user group as I cant see anything on the website..

After playing with Progress in a VM, I think I understand a bit more how it all works. My plan at this stage is to:

  • Script a shutdown of the database, perform a mirror of the DB file with DoubleTake and disable replication. After the mirror completes, mark the database as backed up with rfutil, empty the AI files and restart the databases.
  • Along side this, continuous replication of the AI files to the offsite server.
This should mean that in the event of failure, we have a consistent database backup and the related AI files to the point of failure, whilst keeping WAN traffic down.

Should this work? Do we need the BI files at the point of failure, or can these be snapshotted while the DB is offline?
 
When you snapshot (with it shutdown) the db take the bi files, the ai files, all of the data extents and the .db file.

That will be a valid backup that you can then roll-forward accumulated ai files against.

But don't take my word for it. Test it. Make sure that it works ahead of time and it will work when you need it to.
 
Back
Top