none
S3 database choking during web job runs - what are my options? RRS feed

  • Question

  • We run an set of web jobs that is very intensive on the db for approx. 4-6 hours every day.  Our average DTU load on our S3 db is 21%, however during the web job runs, DTU and CPU are pinned @ 100% load.  This makes our app virtually un-useable during this period.  What are our options in order to still run the web jobs without competing with app performance?

    Is there a method where we can scale up our db on demand to meet the load?



    • Edited by listnet Thursday, April 28, 2016 5:03 PM added image
    Thursday, April 28, 2016 4:57 PM

All replies

  • It seems like your workload is spikey and needs more DTUs to handle the load during spikes. One obvious option is to scale up to P1 but that also means you will have a SKU that isn't being used to the optimum except at the time of spike in load.

    Is this the only DB you have or do you have other DBs for your app? If you have more than one DB, you could look at using elastic pools, which is a great way to share resources between DBs that don't need them consistently. You can read more here : https://azure.microsoft.com/en-us/documentation/articles/sql-database-elastic-pool/.

    Friday, April 29, 2016 5:17 PM
  • We use: 

    S3 x 1 (spike loads)

    S0 x 2 (development)

    Basic x 2 (development)

    Would there be a cost advantage also to using a pool

    • Edited by listnet Friday, April 29, 2016 6:07 PM
    Friday, April 29, 2016 5:29 PM
  • A standard elastic DB pool is slightly more expensive than what you have here in $US.   Another option would be to optmise the queries that are causing the spikes. 

    https://azure.microsoft.com/en-us/documentation/articles/sql-database-performance-guidance/

    https://azure.microsoft.com/en-us/documentation/articles/sql-database-query-performance/

     
    Tuesday, May 10, 2016 3:40 PM