Forum Discussion

packet_rat_7271's avatar
packet_rat_7271
Icon for Nimbostratus rankNimbostratus
May 10, 2017

Severe performance degradation running ElasticSearch/Kibana behind BIG IP (VE) via Portal Access

RFI: We're seeing severe performance degradation running ElasticSearch/Kibana behind BIG IP (VE) [Best-25MB] on AWS using Portal Access model (Version 13.0) . F5 Instance was built using the CloudFormation 3NIC Template. The Elastic Search Cluster and Kibana (Version 5.3) are deployed on the F5 Internal Network/VLAN. The Portal Access configuration was deployed using the F5 Portal Access Setup Wizard. Performance degradation is an order of magnitude over mirror image Elastic Search Cluster and Kibana accessed directly on AWS (sans F5). We have had a case open since April 6th (Service Request C2403209). QUESTIONS: (1) Is anyone else successfully running ElasticSearch/Kibana V5.x behind AWS BIG IP (VE) in Portal Access mode? (2) Has anyone seen similar severe performance degradations when deploying Web Services behind AWS BIG IP (VE) in Portal Access mode?

 

5 Replies

  • We're still not making any progress so tickling this post in hopes someone knowledgable may see same and respond. There's gotta be some folks running Elastic Stack and Kibana Version 5.x behind F5 BIG-IP VE out there?

     

  • I haven't run that behind portal mode yet, but make sure you have the following profiles applied to your virtual server:

     

    • Web Acceleration (optimized-caching)
    • HTTP Compression

    Without those profiles, you'll likely be constantly rewriting the same JS files over and over. Also, after rewritten the content, the Portal Engine does not compress any of the rewritten files, so the compression profile is a good idea.

     

    Typically adding those profiles to your APM virtuals using Portal Access will see a huge improvement.

     

    HTH

     

  • Excellent! Many thanks for the response. We'll give these a try and let everyone know.

     

  • Changed "wan-optimized-compression" to "httpcompression" (Web Acceleration was already set to optimized-caching

     

    No difference -- Portal Access method still takes 300% longer.

     

    Many thanks for engaging though. Keep those ideas coming!!!

     

  • Have you used something like Chrome Devtools to see which requests are slowing things down? Should confirm if it is indeed JS files or possibly XHR requests, etc. That's the next step I'd take.