locked
Where did Azure Library for Lucene.Net go? RRS feed

Answers

  • http://code.msdn.microsoft.com/azuredirectory is the correct URL, as before. As far as I know, it was just a glitch that caused it to disappear. Once it was brought to the attention of the project owner, he republished to get it to appear again.
    • Proposed as answer by Steve Marx Saturday, February 19, 2011 5:50 PM
    • Marked as answer by Steve Marx Saturday, February 19, 2011 5:50 PM
    Saturday, February 19, 2011 5:50 PM

All replies

  • Hi ChrisLaMont,

    Thanks for reporting this issue, I will consult my colleague and return to you as soon as possible.

    Thanks,


    Mog Liang
    Please mark the replies as answers if they help or unmark if not.
    If you have any feedback about my replies, please contact msdnmg@microsoft.com.
    Microsoft One Code Framework
    Monday, February 7, 2011 8:12 AM
  • I'm wondering the same thing.
    Tuesday, February 8, 2011 12:22 AM
  • We do not support Lucene Library for Windows Azure that why the following code is removed from codeplex.

    http://code.msdn.microsoft.com/AzureDirectory/Wiki/View.aspx?title=Home

    As of now we don’t have any solution to use Lucene with Windows Azure however a few customer were able to do the following:

    1.        It is possible to use lucene.net library ( http://lucene.apache.org/lucene.net/) in Windows Azure worker Role and get it working. I haven’t tried it however it seems possible


    bill boyce
    • Proposed as answer by billb08 - MSFT Friday, February 11, 2011 12:45 PM
    Friday, February 11, 2011 12:45 PM
  • Hi there!!

     

    Last night a colleague was able to work with the code developer and the team decided to release the BETA library for Lucene for Azure.

    http://beta.code.msdn.microsoft.com/Azure-Library-for-83562538

    FYI: It is not supported product as it is BETA however you can try it


    bill boyce
    • Proposed as answer by billb08 - MSFT Friday, February 11, 2011 4:47 PM
    • Unproposed as answer by Steve Marx Saturday, February 19, 2011 5:50 PM
    Friday, February 11, 2011 4:47 PM
  • http://code.msdn.microsoft.com/azuredirectory is the correct URL, as before. As far as I know, it was just a glitch that caused it to disappear. Once it was brought to the attention of the project owner, he republished to get it to appear again.
    • Proposed as answer by Steve Marx Saturday, February 19, 2011 5:50 PM
    • Marked as answer by Steve Marx Saturday, February 19, 2011 5:50 PM
    Saturday, February 19, 2011 5:50 PM
  • I have a version of this running live right now on Azure. It implements an architecture very similar to the proposed one in the documentation related to AzureDirectory. It does work - quite nicely in fact, save for one critical issue: since only one indexer can be active on an index at any one time, it's not clear how one can scale the indexer worker role to handle very large numbers of either indices or lots of throughput to the index.  The worker role is a singleton, thereby invalidating the SLA as well. 

    I've been trying to find a good way of segmenting a collection of indices across multiple worker roles through some sort of affinity. This would allow scaling but not increase robustness nor performance of any individual index. That would remain an unsolved issue, as far as I know.  I haven't yet found something practical for the affinity either (given that the worker role instances are initially identical and so would either each grab the same indexes or getting into a race condition). 

     

     

     

     

    Tuesday, March 1, 2011 6:33 PM
  • @Christopher Cardninal

    Have you tried using a named location in Blob storage to act as a gate to eliminate the race condition and balance out the load?

    Here is an idea that will work for bootstrapping the service:

    - Since each Azure role instance has a unique identifier, you may be able to use this to distribute the load programatically.  I was thinking of hashing and sorting the IDs.  I'd assign the first 2 IDs to indexes 0 to 100.  The next 2 IDs to indexes 101 to 200.  So scale out a set of indexes I could assign 10 workers to indexes 101 to 200.

    A combination of Azure Table and Blob technologies may be useful in building this solution.  I'm considering that each worker will need to poll the Azure table to determine if it needs to service a different set of indexes.  You may want to check out a Fluent implementation of a worker polling agent here: http://wag.codeplex.com/

    Hope this was of some help; I'm too busy to code it myself, but hope someone does!

    Wednesday, March 2, 2011 5:59 AM