We use a third party ASP application on IIS5/MDAC 2.7/Oracle 8.1.7 which
has recently begun periodically bogging down.

These periods are marked by all of the following conditions:

- Significant slowdown on the browser interface to the app
- Queuing of ASP requests (via perfmon)
- Significant increase in ASP request execution time (via perfmon)
- Dramatic increase in the rate of requests to the Oracle listener, 10 or
20 per second from the webserver, even though there are never more than a
half dozen actual sessions on the database. Under normal conditions, we
see one request every 5 seconds or more.

It does not require any particular load on the webserver to create this
load, though we cannot create it at will. It appears to occur after a
reboot of the webserver while there are active sessions.

If I clear all of the users off the application and only logon a single
session (to the ASP app) once this behavior begins occuring, the same
thing occurs: we see dozens of connections attempted through the listener
log (and netstat on the IIS box), but only a connection or two on the
database server.

If I bounce IIS/WWW and bring them right back up, the problem sometimes
persists, sometimes clears up. If I reboot the server and bring it right
back up, the problem sometimes persists, sometimes clears. If I leave the
server or IIS/WWW down for a few minutes before restarting, this always
clears the problem.

The vendor is no help.

Has anyone ever seen this sort of behavior on IIS/Oracle?

--
LarryD