Hi Buck,
Actually, I was already involved with it as early as March 2004 . I was
involved in the beta testing as well when it was still named DB2 Information
Integrator. In terms of using setting it up, the first moth understanding
the concepts took a bit of pain but after that, setting it up was relatively
easy. I would recommend going through a DB2 II introduction course which I
believe is still no charge under IBM's ISV program (if you are not an ISV
and there is space, there is a big chance you can get in this course). If
you attend the course, they will gladly give you the lab settings but you
have to get DB2 and DB2 II (WS II), there are 120 day test codes at IBM's
site. After that, invest in about three days of onsite consulting to have
someone come in to just show how federation, replication, setting things up
are done.
Our approach to using DB2 and WS II in consolidating our heterogeneous
databases were teo prong. For data sources below 5 million records, we just
federate accross different databases, meaning data does not leave where they
are, we just define nicknames in DB2 and just create views in case we need
to federate data from different sources. So I create a view from a bunch of
nicknames coming from Oracle, SQL, DB2 and then our reporting tool just hits
these views. For DB2 data sources bigger that 5 million records, we use Data
Propagator to replicatate a subset of this to DB2, SQL, Oracle
appropriately. We found that federation is fast and efficient only up to a
certain volume of transactions, after that performance is an issue, so we
use Data Propagator or Q Replication to distribute data out and re-federate
them back to DB2 for our reporting tool to hit.
In terms of supporting it, it is still true for us that anything IBM when it
is working is great but when it breaks, it is too tough to fix things.
Having said that, with DB2 II (or WS II) in time, you will get familiar with
its quirks. For federation, in two months, one can be a federation expert.
With replication issues especially if you are using replication to massage
data, change values, implement business rules, it may be a bit of a
challenge but if you compare using Data Propagator or Q replication or Web
Publish as part of an ETL (Extact transform and Load) tool, I will still
say, it is easier to use than Informatica or Asenctial. You get more real
time replication as well through Data Propagator or the now faster Q
replication (but you also need an MQ support person if you plan to use Q
replication).
Our approach in having support here is to have a dedicated Replication /
Federation expert. this expert need to have only basic DBA skills (such as
designing tables, executing DML, cleansing duplicate data, but highly
skilled on jdbc drivers, odbc drivers, message queues, database
connectivity. As this person gets used to the product, the support starts to
become easier (6 months heavy exposure will do).
Hope this helps.
RdR
"Buck Nuggets" <bu*********@yahoo.com> wrote in message
news:11**********************@l41g2000cwc.googlegr oups.com...
What I can say is all Databases are the same in terms of pricing,
ease of
use, ease of technical support, being good or bad on it comes at
different times for the different vendors of those companies. I have personally
experienced these for all companies. My conclusion is, whatever
database you use, your data will always be delivered the way you planned and
implemented
your database system.
Well, I've got to respectively disagree with you there: sql server is
still the easiest to manage and develop for in my opinion, but stuck on
windows, lacks partitioning, lacks decent exception handling in its
stored procedures, is crippled from the command line - and if you
actually want to promote DTS packages from dev to test to prod it's
incredibly difficult. Oracle has a fabulous partitioning solution -
great for real-time loading, great 3rd-party support, etc. But is the
most expensive, and most complex to manage. DB2 has improved usability
& online maintenance tremendously in the last few years, and has a
price & ease of maintenance benefit over oracle in my experience now.
Still plenty of areas in which one has the edge over the other - a nice
competitive arrangement I think. :-)
Anyhow, my question is - have you worked with Websphere Information
Integrator? I thought it just became named that this year...
Are you actually federating data across multiple vendor products with
it?
Was it easy? How long did it take you to set it up, learn the product,
and slip it into production?
Do you think it would make federating a dozen db2 databases easier than
just doing it manually with nicknames & replication (if you must)?
When it breaks do you need a team of db2, oracle, j2ee, ii, and
webspere experts all in the same room to make sense out of it?
Thanks!
Buck
