[jira] [Updated] (USERGRID-367) Refactor entity writes to run concurrently

["Todd Nine (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/USERGRID-367?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Todd Nine updated USERGRID-367:
-------------------------------
    Description: 
We need to implement our operations concurrently to increase our throughput.  We need to execute the following operations in parallel.

Entity Write
Graph Edge Write
Entity Indexing (still async)

We have lazy repair on all reads if the system were to get out of sync.  Since we do not have a transactional store, we can utilize this read repair to increase write throughput.  In the event a transaction fails, we attempt to roll back. If the roll back fails, it will be repaired the next time a stale entry is read.


> Refactor entity writes to run concurrently
> ------------------------------------------
>
>                 Key: USERGRID-367
>                 URL: https://issues.apache.org/jira/browse/USERGRID-367
>             Project: Usergrid
>          Issue Type: Story
>            Reporter: Todd Nine
>
> We need to implement our operations concurrently to increase our throughput.  We need to execute the following operations in parallel.
> Entity Write
> Graph Edge Write
> Entity Indexing (still async)
> We have lazy repair on all reads if the system were to get out of sync.  Since we do not have a transactional store, we can utilize this read repair to increase write throughput.  In the event a transaction fails, we attempt to roll back. If the roll back fails, it will be repaired the next time a stale entry is read.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)