Testing and optimizing Web Performance on the EPiServer platform Page 1
We test, analyze, monitor & improve Web Performance Our vision Page 2
Sample Customers Public sector Finance Telecom Media Travel Gaming Other Page 3
The 3 Web Performance challenges 1. How many concurrent user can I handle? 2. Where are my bottlenecks? 3. How can I avoid overload? Page 4
EPiServer is an excellent base for excellent performance 4 5 6 4 5 6 4 5 6 4 5 6 4 5 6 Media Travel Corporate E- Commerce Community Page 5
Hierachy of Performance Infrastructure Network FW Load balancers Web Servers IIS / Windows DB Server SQL Server EPi-server Web Application Startpage Content Application Login Logic Transactions Page 6
How many concurrent users can I handle? How many users with good response times When does it break? Page 7
Capacity Cost IT vs. Marketing Income Load 9 June, 2010 Strictly confidential Page 8 Page 8
Step 1 Setting Performance Targets Examples We shall never crash due to load We shall handle100.000 page views per hour Better then 4 sec average response time We shall compareble with the best in class sites for... Our peak time response time shall be better then competition s 95% of our users shall make a travel selection for purchase in less then 30 seconds Page 9
Step 2 - Apica Load testing Methodology All applications have bottlenecks no surprise.. There are bottlenecks and Achilles heels.. They can only be effectively tested one at the time Page 10
Page 11
Test Case What is beeing tested How it is beeing tested How it is analysed What is the recommendation Select stock Stop or GO : Next iteration Page 12
Typical Load test results Page 13
2 sides of the Coin 1. Throughput How much can you deliver? 2. Concurrency How many at the same time? Page 14
Stress test vs concurrent users Page 15
The 3 Web Performance challenges 1. How many concurrent user can I handle? 2. Where are my bottlenecks? 3. How can I avoid overload? Page 16
Test Details Page 17
Typical Report 1 2 Before and after Plugin : ImageVault default ( red ) Quick Time locally at the servers ( Blue) Page 18
Find and fix the weak points 1. Limited bandwith Local Hosting Burst Option Or CDN / Apica WebOverload > 100 Mb/s 2. Limited WEB Server CPU Front end Cache / Mem Cache EPi Multiplexing of TCP/ IP Session Netscaler / BIG-IP or Apica WebOverload 3. Fixed transaction speed How many transactions per minut What if that is exceeded? Queue based on Speed 4. No limit of concurrent users New customers crashes the application Queue based on Sessions > 80% < 10 Mb/s < 10 000 Page 19
The 3 Web Performance challenges 1. How many concurrent user can I handle? 2. Where are my bottlenecks? 3. How can I avoid overload? Page 20
Expect the unexpected... Page 21
Apica WebOverload Customer Local Servers 1.DNS Customer 2.DNS redirect to APICA GSLB 1 1 2 2 WAIT GB-lina Apica Cloud Page 22
Dynamic queue management All applications have a max High load above max capacity often leads to a complete stop Solution - dynamic queue management Rerouted to a static Wait page Queued users will not destroy Snowball effect Page 23
Wait Page Page 24
Performance improvement in three steps Testing the site s or service s baseline capacity Optimizing response times and baseline capacity Protection against overload when needed Page 25
Partner benefits from LoadTesting Proactive load testing detects and attributes performance bottlenecks to the proper part/component in your project Your customer will enjoy a faster web site Your customer will enjoy a more more stable web site EPi Testing Index & Best practise Page 26
Get started on a free trial Special offer for EPiServer Partner Summit attendees Apica LoadTest - One free test of up tp 100 simultaneous users per minute Apica WebPerformance - 1 Full Page render check and 4 basic URL checks for 60 days Go to epi.apicasystem.com Thank you! Page 27