locked
Local Development Fabric Storage Slow RRS feed

  • Question

  • Does anyone know why reads and writes from local development fabric blob storage takes anywhere from 300ms to 1200ms per request?  This is running on my local machine -- why is it so slow?

    Cheers,
    Shaun
    Tuesday, March 16, 2010 7:04 PM

All replies

  • Am I the only person who experiences this?  Why would a request to pull a 5KB blob from localhost to localhost take longer than a second?  It's actually faster for me to access real cloud blobs than it is to access them locally.
    Friday, May 14, 2010 4:16 PM
  • I've got a project where I'm uploading 100,000+ blobs (~1.3kb each) on startup but it is taking several hours to complete for what should be a simple SQL INSERT (for development storage), not to mention DSService.exe is using 75-85% of my CPU for the entire process.
    Friday, May 21, 2010 2:33 PM
  • This seems very suspect.  I'm having the same problems -- I can't imagine why it is so slow to access the fabric locally.
    Friday, May 21, 2010 2:59 PM
  • I'm also having a lot of problems with local storage, specially regarding to its performance. Actually, it takes more than 60s to write a 16.7kb blob to it, where it takes only a few ms to write the same file to the production storage.

    Friday, September 24, 2010 9:07 PM
  • Based on my understanding, Azure storage is not SQL server based like DevStorage is. So it might be quite possible to have difference in performances betweeb DevStorage and Azure Storage.
    Please mark it as answer by clicking on "Propose As Answer", if it helps
    • Marked as answer by Brad Calder Sunday, March 27, 2011 12:58 AM
    • Unmarked as answer by S Tonstad Wednesday, June 1, 2011 5:08 PM
    Saturday, September 25, 2010 12:13 AM
  • This is still a huge problem. I can't even delete files from development blob storage that are bigger than about 50 MB because the operation just times out.
    Tuesday, April 12, 2011 10:42 PM
  • It remains bugged in the 1.4 SDK release.  For any kind of high I/O activity with development storage requests just get dropped.
    Wednesday, June 1, 2011 5:08 PM
  • I'm having the same issue. At first I thought it was a threading issue, but having a single thread attempt to write both a queued message and a blob to dev storage takes upwards of 2 seconds per pair.

    It's not a show stopper for me, but it makes some types of tests difficult.

    Tuesday, July 26, 2011 7:59 PM
  • Would be helpful for Microsoft to address this issue.  Either acknowledge it or discuss plans for fixing the developer story for local storage.  It is abysmal in current form.

    Tuesday, July 26, 2011 8:21 PM