Answered Publish a service without interruption

  • 2012년 8월 1일 수요일 오후 12:55
     
     

    Hi everyone, do you know if it is possible to update an azure service composed of a Worker Role and Web Role without any interuption? because usually when we publish the actual service get stopped and the new one start getting up, but this for us is a problem because our customer need the webservice up and running, a delay of a few seconds is not a problem, but this usually can take 10 to 15 minutes.

    There is a way to solve this?


    Touch the future with iBox. http://www.ibox-media.com/

모든 응답

  • 2012년 8월 1일 수요일 오후 1:27
     
     답변됨

    Hi,

    Usually inplace update for service roles wont cause any down time unless you change the machine size. But you can safely deploy your applications with out downtime if your deployment is only specific to application changes.

    I recommend you deploy the changes first to staging and make sure that they are working and do a VIP Swap which wont cause any down time for your applications and automtically traffic would be routed to new deployments in production. But in this approach you must make sure that the number endpoints are same for both staging and production versions. any change to the number endpoints will result in failure for VIP Swap.

    Please refer to this document that helps you to chose the best approach based on your requirement.

    If you let me know more details of specific issues i would try to answer them.



    Please mark the replies as Answered if they help and Vote if you found them helpful.

    • 답변으로 제안됨 Veerendra Kumar 2012년 8월 1일 수요일 오후 3:10
    • 답변으로 표시됨 Orangebit 2012년 8월 1일 수요일 오후 3:14
    •  
  • 2012년 8월 1일 수요일 오후 2:58
     
     

    Hi,

    Usually inplace update for service roles wont cause any down time unless you change the machine size. But you can safely deploy your applications with out downtime if your deployment is only specific to application changes.

    I recommend you deploy the changes first to staging and make sure that they are working and do a VIP Swap which wont cause any down time for your applications and automtically traffic would be routed to new deployments in production. But in this approach you must make sure that the number endpoints are same for both staging and production versions. any change to the number endpoints will result in failure for VIP Swap.

    Please refer to this document that helps you to chose the best approach based on your requirement.

    If you let me know more details of specific issues i would try to answer them.



    Please mark the replies as Answered if they help and Vote if you found them helpful.

     ok i will try by using two instances, because i have a worker role that is doing background data processing, but is not so important that is always running, but i have a WebRole with a webservice that i need it to be always running.

    I will try to add an instance to the webrole and see if with this change i can keep the webservice always running until the other instance is updated.


    Touch the future with iBox. http://www.ibox-media.com/

  • 2012년 8월 1일 수요일 오후 3:10
     
     
    Yes, thats the right approach to be covered under SLA agreements and maintainence of application in a better way.


    Please mark the replies as Answered if they help and Vote if you found them helpful.