Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

My organisation currently has 2 data centres from which it hosts its corporate website, with Akamai providing load-balancing, and other CDN services. Now, I'm told that if our server stack in 1 data centre stops responding, Akamai will automatically route the incoming traffic to the alternate data centre.

Q: Now my question is, what is the trigger that causes Akamai to initiate this routing?

ie. Does Apache Http server literally have to stop listening/responding on port 80, or would returning non-2XX/3XX/4XX codes do it? (ie. have the application server return HTTP 500's)

If so, can it be configured to work on a context-by-context basis, so that if http://www.example.com/search goes down (app server returns HTTP 500), traffic to /search is routed elsewhere, but all other traffic still hits both stacks?

Q: Also, how long after restoration of service would traffic start being routed back?

share|improve this question
    
You should ask Akamai about all this stuff. –  gekkz Apr 19 '12 at 16:04
    
Indeed I should, however the only person registered as an authorized contact got on a plane and is half way around the world. Will be out of contact until Monday. Nvm. –  Crollster Apr 19 '12 at 16:48
add comment

1 Answer

up vote 3 down vote accepted

(1) This is one the fail-over, typically called Fail-over to Alternate Data Center "Mirror Fail-over"·

Fail-over can be directed to use the backup site in case the primary is unavailable. You may choose to have the backup site be different from the origin site - for example, it may contain reduced functionality or content. Upon receiving a request for content that resides on the primary site and determining that the primary site is unavailable, the Akami edge server will obtain the request content[ or its defined substitute] from the mirror site "Alternative DC" in a manner invisible to the end user.·

(2) The restoration depends on your thresholds. What is your frequency of your check and how long your response need to be validated. When I was using Akamai. I add some extra checks to ensure the DC restoration...like

  • Edge bandwidth usage drops or bursts
  • Does Origin server within a data center fails.·
  • Is there a Origin connection failure "Associating to DC load balancer"·
  • Origin DNS failure·
share|improve this answer
    
Thanks for your answer. Let me see what options I have access to now that I can contact Akamai. –  Crollster Apr 25 '12 at 8:35
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.