Very high latency in fetching static files

Question

Problem : Really high latency in fetching static files upon first request.
Environment : Weblogic V8.1 App Server and Web Server running using jdk1.4.2_13
Mem Setting: -Xmx2048M -XX:MaxPermSize=1024M
Situaton: Restart the weblogic server, fetch a static file using " wget hostname:port/script.js". Comes back in around 100 ms for us. Then from browser we request "http://hostname:portname/Logon.do". Then again we do " wget hostname:port/script.js" and now it comes back in 400 ms. The above times are in our TEST environment. In PROD it is really worse 1300ms for one file !!!
Tests:
Wrote a servlet to handle all static file (CSS,js,gif,jpeg etc) and all requests would go through this servlet. Did not solve the problem, latency is still high.
Wrote a test servlet to load about 1000 classes explicitly (Class.forName) and then fetched the static file with out accessing Logon.do from browser. Latency was high after loading classes. This lead to conclusion that it is number of classes we are loading which is causing the problem.
Ran application under JProfiler, Took a heap dump right after starting server. About 1300 classes loaded. Took a heap dump after fetching Logon.do. About 2100 classes loaded.
Question :
1) Isn't 1300 too high for just the server to start? (there are some classes loaded because of our references in ejb's)
2) We use prototype factory to instantiate classes. All the interface and implementation definitions are specified in config file and it seems to load all the classes. May be we can change this. But what confuses me is that the response time was okay when we had 1300 classes loaded how come it became so bad when there were 2100 classes ?
If any body is using these same set of technologies, what is the average number of classes loaded on initial start up?
I am running out of options to test?

Answer 1

There's no way from what you posted that you could draw a conclusion that you are infected.
I totally missed that tracert, but from a ping response time, it looks healthy.
The packet loss, could easily cause your problem.
Does it consistently loose packets there ? Cause I've pinged it from here and got no loss.
Also remember that in this case, ping is also the time it takes the blizzard servers to response to your network message, so it could be local to blizzard, and ping might not detect that.
Have you tried a netstat -p tcp -b command too see whats active on your PC ? That could point you towards something local.