Stuart Boydell wrote, " I'm wondering what peoples experience of using
UniOLEDB is.".  Here is our experience.

Our environment experience has been on version 6 of UniData on Windows
2003.

1)  Our engineer's had a heck of a time getting the right connection
string to work.  It has been such a long time back that I don't remember
what the problems seemed to revolve around.  Once it started we moved
on.
2)  We used it for an ASP.NET application.  I'm not a .NET engineer but
since it is a COM DLL I'm certain there was COMInterop involved.  Since
it was COM then yes it ran in unmanaged code.  We were able to implement
C# code that called interfaced with the DLL and ran as a Windows
service, getting its request from a MSMQ queue.  We did not use ADO.NET
pooling.  I would have to query and engineer on why, but my guess is it
does not fit.
3)  The real issue is it uses the UniData SQL engine, which did not
perform to our expectations.  Additionally we implemented our own
connection pooling by using window services and MSMQ queues, but we were
told this is not how it is intended to be used.  It is intended more for
a client server implementation.  Either as a result of our
implementation or the UniOLEDB implementation, we seemed to have issues
with timeouts that fully could not be explained.  The overhead of using
VSG to define the SQL Schema and setting up uci.config and ud_database
was not appealing.
4)  Note we only used it for querying data and not updates.  We have a
lot of business logic tied up in UniBasic and SB+ therefore we have to
use UniObjects (and RedBack in the past) to call subroutines.  So you
end up having more database connections than you would like.
5)  Our impression has always been that having UniData on Unix on the
backend will perform better and be more stable.  Our problem is the
majority of our market will not support Unix deployments whether on
principal or practice.

We are running this environment in production though and it is workable
just too many issues.  Therefore we are moving away from it and using
UniObjects.NET and UniData 7.1.  We are switching all our SQL statements
over to UniQuery statements that generate data to XML.  So for most of
the simply queries with string data have worked within the MS dataset.
We are using XSLT to for the others.  One reason is to get in an
environment IBM supports and that is not a COM implementation.
Additionally IBM says that native access is always faster (native being
subroutines and uniquery).

If you have questions you can contact me directly.  Also though I have
not had any experience with it you may want to look at MV.net.

Glenn Batson
360.256.4400
www.jenkon.com
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to