RetrieveMultiple performance on large datasets

EDIT (21/05/15): After DMing with @maustinjones, I have added stat for paged RetrieveMultiple as well. I have updated the post to include this.

EDIT (22/05/15): Updated post to reflect correct behaviour of no-lock when parallelising. Thanks @maustinjones. Please also refer to the follow up post -> http://dreamingincrm.com/2015/05/22/redux-retrievemultiple-on-large-datasets/ on why paging cookie, is the recommended approach.

MSCRM limits the maximum result size to 5000 records, when you use the RetrieveMultiple request. In order to overcome this and retrieve all the records, you’ll have to use paging. PFEDynamics team have also released an open-source library called PFE Xrm Core Library that utilises Task Parallel Library.

There is also ExecuteMultipleRequest that you can use to send bunch of requests in one go, and process the responses. This just wanted to document by findings, about the performance of these options:

1. Just using Parallel.ForEach
2. ExecuteMultipleRequest
3. PFE Xrm Core
4. Paged RetrieveMultiple

Run 1 (Batch…