ignite-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexei Scherbakov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (IGNITE-4523) Allow distributed SQL query execution over explicit set of partitions
Date Sat, 07 Jan 2017 14:11:58 GMT

    [ https://issues.apache.org/jira/browse/IGNITE-4523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15807556#comment-15807556
] 

Alexei Scherbakov commented on IGNITE-4523:
-------------------------------------------

I implemented query execution on nodes assigned as primary for partitions from a set passed
to a query.

Implemented a test for the scenario mentioned in previous post: partitions are divided to
subsets and each subset is bound to a some region, on example, geographic.

So we have data in different regions and are able to do distributed queries over a region
without broadcasting to whole cluster and skip index definition on region.

Need to add more tests before making PR.

> Allow distributed SQL query execution over explicit set of partitions
> ---------------------------------------------------------------------
>
>                 Key: IGNITE-4523
>                 URL: https://issues.apache.org/jira/browse/IGNITE-4523
>             Project: Ignite
>          Issue Type: Improvement
>          Components: cache, SQL
>    Affects Versions: 1.8
>            Reporter: Alexei Scherbakov
>            Assignee: Alexei Scherbakov
>             Fix For: 2.0
>
>
> Currently distributed SQL query is executed on all nodes containing primary partitions
for a cache, sending map query requests on all nodes in grid.
> Sometimes we know in advance which partitions hold a data for query, on example, in case
of custom affinity function. 
> Therefore it's possible to reduce number of nodes receiving map query request by providing
explicit set of partitions, which will give significant performance advantage and traffic
reduction in case of very large clusters.
> Internally we already have such functionality, so the only necessary thing is to provide
public API for what.



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

Mime
View raw message