Skip to Main Content

Java APIs

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

Interested in getting your voice heard by members of the Developer Marketing team at Oracle? Check out this post for AppDev or this post for AI focus group information.

Latency issues with TCP Acknowledge

843790Mar 11 2008 — edited Mar 12 2008
We have a very bewildering problem that I feel should be in this fourm, if not, please move it to the correct forum.

Currently we have Java Web Services that are deployed in two different enviroments, one in Windows, one in Linux (both deployed via Tomcat). We also have two corresponding destinations (one in Tomcat, one in IIS) for those services (publish/subscribe). Now here is where it gets dicey.

All of the tests are run on separate boxes

With the server running on Unbuntu(Tomcat) and the destination running on Unbuntu (Tomcat) the problem does not show.
With the Server running on Ubuntu(Tomcat) and the destination running on Windows(IIS destination with .Net) the problem does not show.
With the server running on Windows(Tomcat) and the destination running on Windows (Tomcat) the problem does not show.
With the server running on Windows(Tomcat) and the destination running on Windows IIS destination with .Net) the problem shows.


The problem is as follows.

The server sends the initial payload to the client, the client sends back a HTTP Continue, and then the server sends back an ACK for the HTTP Continue. The problem is, that it takes almost 2 tenths of a second for the server to send the ACK, where as all of the other configurations send the ACK in less than one hundred thousandth of a millisecond.

What is baffling, is the server running on Ubuntu Server handles the ACK to the IIS destination nicely, but the Windows Server takes almost 2 tenths of a second.

The other difference is in the TCP packet coming back to the server there is a difference of 12 bytes from the Tomcat destination to the IIS destination, which is the addition of the Options field.

The server is using Jax-WS for stubs, BTW, even though I feel this is down pretty low in the HTTPClient or lower.

Any ideas?

Comments

Processing
Locked Post
New comments cannot be posted to this locked post.

Post Details

Locked on Apr 9 2008
Added on Mar 11 2008
4 comments
398 views