[jira] [Updated] (IGNITE-5937) Mvcc data structure for SQL queries

["Vladimir Ozerov (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IGNITE-5937?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Vladimir Ozerov updated IGNITE-5937:
------------------------------------
    Component/s:     (was: sql)

> Mvcc data structure for SQL queries
> -----------------------------------
>
>                 Key: IGNITE-5937
>                 URL: https://issues.apache.org/jira/browse/IGNITE-5937
>             Project: Ignite
>          Issue Type: Sub-task
>          Components: cache
>            Reporter: Semen Boikov
>             Fix For: 2.2
>
>
> Need implement some data structure to store/query multiple entry versions.
> One possible option:
> - committed value at first is stored in separate BPlusTree index (there also need store related tx id to filter out data for non-finished transactions)
> - periodically flush data for finished transaction in 'main' index
> - for SQL queries need merge result from main index and filtered 'mvcc' index. Note: it is possible 'mvcc' index will contain multiple committed versions of the same entry, need make sure only one last one will appear in result.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)