Twice now, I've been involved in a DBMS selection for a large
organization. It is not a trivial exercise. Nor is there a single way
to rate the products.
I would suggest that you start with listing your organizational
requirements. You might want to start with things such as,
What will the DBMS be primarily used for (ie. OLTP, DDS, object
repository)?
What volumetrics can you forsee (ie. number of concurrent users, data
volumes)?
What hardware and OS will you use?
Is high or low end scalability a requirement?
Do you expect to run the DBMS on multiple platforms?
What is/will be your data architecture (ie. centrallized, fragmented,
replicated, etc)?
What front end tools, systems management and monitoring tools do you
expect to leverage into the new dbms?
Do you have licencing and support advantages, based on your current
technical environment, to go with one vendor or another?
The list goes on...
Largely the big DBMS manufacturers all produce good products (or bad
products, depends who you talk to). A fair bit of leapfrogging goes on
as one vendor or another puts out a new version. Largely disregard
that.
Focus on:
Fit within you organization,
Whether the DBMS is available on the hardware and OSs you will use,
and whether these are strategic platforms for the vendor (as opposed
to being released late in the DBMS life cycle - ie. Sybase on O/S2),
Whether the DBMS is interoperable with other product you have (ie, TP
monitor, backup software, a financial package you use, security
package, ect).
Whether it scales well for your needs,
Whether it has the niche features you may need (ie. replication,
bitmapped indexes, OLAP, OO, etc),
Whether you neck of the woods has enough DBAs in the particular DBMS
to support the product over the long term,
Whether the vendor has local presence,
Viability of the vendor/product,
Vendor support track record,
Third party tools availability,
Systems monitoring/management tools available,
Cost.
etc.
Performance is important, but one that keeps improving for all
vendors. Leapfrogging keeps changing the picture, but consider the big
DBMSs roughly equivalent. It's more an issue in sizing a particular
configuration than in choosing a DBMS. You will want to look into
performance, but unless things are really out of whack don't base
selection on it.
DBMS selection is not trivilal. You may wish to bring in an expert to
look into this. The DBMS will, likely, become a cornerstone of you
organization that you will have to work with for many years. The long
term investment warrants taking a good look at what's out there.
Good luck,