Bad Performance In a Web Cache Cluster When The Network Interface Cards Of One Of The Members Is Marked Down

(Doc ID 1931261.1)

Last updated on NOVEMBER 08, 2016

Applies to:

Portal - Version 11.1.1.1.0 and later
Web Cache - Version 11.1.1.0.0 and later
Information in this document applies to any platform.

Symptoms

In a high availability setup with multiple Portal midtiers, an performance for invalidation requests degrades when the the network interface card (NIC) of one of the web cache cluster members is not reachable anymore on the network. Upon making the NIC availalbe again on the network, perofrmance is restored immediately. Typical scenarios which can lead to the performance degradation are :

The performance degradation can be demonstrated with the Web Cache Invalidation API. When all cluster members are up and running, invalidation requests through the API take no more than miliseconds. Performance of the invalidation requests drops to the second range when the NIC is not avaialble anymore.

Test results :

Invalidation performance when the NIC is up

<?xml version="1.0"?>
<!DOCTYPE INVALIDATIONRESULT SYSTEM "http://www.oracle.com/webcache/90400/WCSinvalidation.dtd">
<INVALIDATIONRESULT VERSION="WCS-1.1">
<SYSTEM>
<SYSTEMINFO NAME="WCS_CACHE_NAME" VALUE="mt1.acme.org-webcache1"/>
<SYSTEMINFO NAME="WCS_NUM_OBJECT" VALUE="1"/>
</SYSTEM>
<OBJECTRESULT>
<BASICSELECTOR URI="http://www.acme.org:80/portal/page/portal/Design_Time_PG/Welcome" />
<RESULT ID="1" STATUS="SUCCESS" NUMINV="1"/>
</OBJECTRESULT>
</INVALIDATIONRESULT>

PL/SQL procedure successfully completed.

Elapsed: 00:00:00.01

Invalidation performance when the NIC is down

<?xml version="1.0"?>
<!DOCTYPE INVALIDATIONRESULT SYSTEM "http://www.oracle.com/webcache/90400/WCSinvalidation.dtd">
<INVALIDATIONRESULT VERSION="WCS-1.1">
<SYSTEM>
<SYSTEMINFO NAME="WCS_CACHE_NAME" VALUE="mt1.acme.org-webcache1"/>
</SYSTEM>
<OBJECTRESULT>
<BASICSELECTOR URI=""/>
<RESULT ID="1" STATUS="Cannot connect to WebCache invalidation port"
NUMINV="0"/>
</OBJECTRESULT>
</INVALIDATIONRESULT>
</INVALIDATIONRESULTDETAIL>

PL/SQL procedure successfully completed.

Elapsed: 00:00:05.69

Symptoms

Environment :

The issue was tested in a six node Portal high availability cluster with Fusion Middlware 11g R1 (11.1.1.7)

Test suite setup

  1. On the Portal midtier server, change directory to $ORACLE_HOME/webcache/toolkit. This directory contains the code for the Web Cache Invalidation API (wxvutil.sql)
  2. From this directory, start SQL*PLUS and connect to the database, preferably with a dummy user such as SCOTT or any other account for testing purposes
  3. Connect as database administrator and grant privileges to the test user to use the network layer from within the database :

     
  4. Repeat the same test with the NIC of one of the cluster members down. The reported elapsed time for the invalidation request will be higher.

 

Cause

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms