Over a million developers have joined DZone.
Platinum Partner

Infinispan performance tweaks

· Performance Zone

The Performance Zone is brought to you in partnership with New Relic. New Relic APM provides constant monitoring of your apps so you don't have to.

 This article is a follow up to Getting started: Infinispan as remote cache cluster

Out of the box Infinispan configuration works great for low to medium number of GET/PUT operations. But in distributed mode and for heavy  GET/PUT operations, you may frequently see locking failures like this one:

2013-03-22 00:14:20,033 [DEBUG] org.infinispan.server.hotrod.HotRodDecoder HotRodClientMaster-63 - Exception caught
org.infinispan.server.hotrod.HotRodException: org.infinispan.util.concurrent.TimeoutException: Unable to acquire lock after [10 seconds] on key [ByteArrayKey{data=ByteArray{size=18, hashCode=48079ac7, array=0x033e0f3134354065..}}] for requestor [Thread[HotRodClientMaster-63,5,main]]! Lock held by [(another thread)]
        at org.infinispan.server.hotrod.HotRodDecoder.createServerException(HotRodDecoder.scala:214)
        at org.infinispan.server.core.AbstractProtocolDecoder.decode(AbstractProtocolDecoder.scala:75)
        at org.infinispan.server.core.AbstractProtocolDecoder.decode(AbstractProtocolDecoder.scala:45)

Infinispan uses locking to maintain cache consistency. Optimizing locking settings can help improve overall performance. Here are some configuration tips to avoid locking issues and improve concurrency:

        <locking concurrencylevel="1000" isolationlevel="READ_COMMITTED" lockacquisitiontimeout="500" uselockstriping="false">
        <jmxstatistics enabled="true" />
        <!-- Configure a asynchronous distributed cache -->
        <clustering mode="distribution">
            <hash numowners="2"></clustering>
  • Concurrency level: Adjust this value according to the number of concurrent threads interacting with Infinispan.
  • lockAcquisitionTimeout: Maximum time to attempt a particular lock acquisition. Set this based on your application needs.
  • useLockStriping: If true, a pool of shared locks is maintained for all entries that need to be locked. Otherwise, a lock is created per entry in the cache. Lock striping helps control memory footprint but may reduce concurrency in the system.
Another configuration worth looking at it Level 1 (L1) cache. An L1 cache prevents unnecessary remote fetching of entries mapped to remote caches by storing them locally for a short time after the first time they are accessed. Read more here.

The Performance Zone is brought to you in partnership with New Relic. New Relic’s SaaS-based Application Performance Monitoring helps you build, deploy, and maintain great web software.


Published at DZone with permission of Nishant Chandra , DZone MVB .

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}