3

Hy,

I am developing an Episerver Site, and experiencing a very poor page load speed

Configuration used:

  • Amazon AWS Windows Server 2008 R2 Datacenter
  • 8 GB RAM, 64 bit op. system
  • Microsoft Sql Server 2008 R2
  • IIS 7.5

Followed http://world.episerver.com/Blogs/Alexander-Haneng/Dates/2011/12/Starting-CMS-6-R2-sites-faster-after-build/ blog post, but not got a huge performance changing

Also enabled Trace in web.config, and i see in trace.axd that between: Begin Load and End Load there is a huge loading time: 20 seconds sometimes, and not for the first load.

Also having trouble when enabling "Dynamic data cache in IIS" which causes the pages to not load any more, i got error message content encoding error, as i saw this comes from Content-Length Header, that it is not matching the required length.

I am not understanding why takes a page to load in ~4 seconds in general, any help would appreciated, thanks

Kszili
  • 334
  • 3
  • 16
  • How does the site perform with the same database in your development enviroment? – Johan Kronberg Mar 01 '13 at 11:58
  • A little faster on my dev env with Windows7, i just noticed that the cpu on a specific page gets to 44% with one request from a client, tested with 2 clients and got 88% CPU usage, very annoying – Kszili Mar 01 '13 at 12:19
  • How many EPi pages are there? How big is the database? – Johan Kronberg Mar 01 '13 at 12:36
  • about 1000 pages, Found something, that DF.GetDescendents() could be the problem, for ex, if i give to get descendents for a page which has 300 sub pages? – Kszili Mar 01 '13 at 12:52
  • Getting a lot of subpages has historically been quite slow in EPiServer. Try removing the call to GetDescendents just to see if that's your bottleneck. – Anders Bornholm Mar 01 '13 at 13:18

2 Answers2

3

Without optimization EPiServer can be quite slow at cold start, but serves pages extremely fast unless you run out of memory in the app pool - and you shouldn't with 8 GB of RAM.

Things to check:

  • App pool behavior: does CPU spike with every page load? Is the app pool using a lot of physical memory? Are there any memory limits imposed on the app pool?
  • On the start page of Admin mode of the site backend there are cache stats. Cache hits should be really high (> 90%) or something is wrong with the cache.
  • Are you sure it's server-side lag - what does Developer Tools/Firebug say?
  • Have you enabled log4net logging? It can be extremely useful when troubleshooting performance issues. If yes, have you mapped your own configuration file? (we had extreme performance issues from that - use the built-in mapping to episerverlog.config)
  • Are you having issues with all templates? What if you create a barebone test template with no master page/main layout view and do nothing but output a property.
  • Any suspicious events logged in the event log?
Anders Bornholm
  • 1,326
  • 7
  • 18
  • I was testing CPU Spikes, commented above for @JohanKronberg , and on an other page, first few times gots high CPU, and after that no more CPU usage (1,2%), And in admin mode the percentage is 99 % – Kszili Mar 01 '13 at 12:21
2

The answer is most likely in your templates if the site is slow in both DEV and Amazon environment.

Try a "barebone test template" as Anders suggests.

Also do a trace using for example dotTrace http://www.jetbrains.com/profiler/ and/or possibly SQL Profiler to see if you find something slow or something that occurs way too often.

Johan Kronberg
  • 1,086
  • 7
  • 12