Db2 mirror tables

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Sprellari - 27 Jun 2006 11:25 GMT

Hi.

I need to mirror a set of my database data, that is create a database
which acts as a buffer in front of the real one.
We have in our DB2 database data which has been created over the last
100 years or so, and is obviously huge amount of data. All normal
operation is done though on data which has been created over the last
30 days or so.

Is it possible to create database which only holds data created over
the last 30 days. Further more all inserts, updates and deletes should
affect both the huge database and also the small one. Any suggestions
would be greatly appriciated.

Reply to this Message

Larry - 27 Jun 2006 13:15 GMT

> Hi.
>
[quoted text clipped - 9 lines]
> affect both the huge database and also the small one. Any suggestions
> would be greatly appriciated.

Well ... this is not what I would call a mirror. Best way to accommodate
this is to do it within your design schema and your application, IMHO.
There are many other questions, though, that must go into this. For
example, how much data is involved in each? Does the "historical" data
ever need to be accessed after it is inserted to? Does it ever need to
be joined with the "30-day" data?

Larry Edelstein

Reply to this Message

Ian - 27 Jun 2006 13:52 GMT

> Hi.
>
[quoted text clipped - 9 lines]
> affect both the huge database and also the small one. Any suggestions
> would be greatly appriciated.

It *might* be possible to do this using an MQT, although I think there
is probably a better solution:

You could look in to using multi-dimensional clustering (MDC), or (even
better), look into range partitioning that's available in DB2 9.

Reply to this Message