Re: Oracle Streams v Advanced Queuing/Replication

Greetings,

Be aware that any kind of replication be it streams, advanced replication, Quest's Shareplex, etc. has a serious learning curve to really get up to speed on the details. If you have the money, go to the classes on the options considered before committing. If you plan to do a true master-to-master, peer-to-peer, active-to-active, whatever you choose to call it, bidirectional environment, the learning curve is much steeper and the opportunities for something to go wrong much greater.

Issues to consider for each option:

(1) Which datatypes do they allow? Which datatypes will you need? Do
you use the XML datatype? LOBs? Raw? Long Raw? Those are some datatypes to ask about.
(2) What limitations will the replication option impose on developers?
(3) What transactional load capacity do you need (if heavy, probably
steer away from AR).
(4) How do they handle triggers?
(5) Will you need a built-in compare/repair functionality (find and
fix synch issues)?
(6) Ask if a DML statement that processes 100 million rows will become
100 million individual inserts on the target. If so, consider the performance hit and potential for falling behind in replication. Do you even care about non-fatal lag-time?

There are lots of other considerations to make in choosing a replication tool, but those are a few good ones to think about. Hope that helps.

neilx.xisenbergx_at_gxmail.coxmx
(remove x's)
Received on Mon Jan 30 2006 - 22:23:28 CST