[
https://issues.apache.org/jira/browse/MRQL-90?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15718841#comment-15718841
]
ASF GitHub Bot commented on MRQL-90:
------------------------------------
Github user fegaras commented on the issue:
https://github.com/apache/incubator-mrql/pull/30
Looks fine but I have to disable tests on Storm mode because some queries
can't pass the tests.
> Add support for Storm streaming
> -------------------------------
>
> Key: MRQL-90
> URL: https://issues.apache.org/jira/browse/MRQL-90
> Project: MRQL
> Issue Type: New Feature
> Components: Streaming
> Affects Versions: 0.9.8
> Reporter: Leonidas Fegaras
> Assignee: Leonidas Fegaras
>
> This task is assigned to Achyut Paudel achyut.paud...@mavs.uta.edu as part of
> his MS thesis at UTA. This extension will allow MRQL queries to run on Apache
> Storm DSPE (distributed stream processing engine). The stream unit is a Tuple
> with just one component: a Bag<MRData>, which is a buffer of MRData used for
> batch streaming. Each MRQL physical operator will correspond to 1-3 Storm
> bolts, while a source operator will correspond to a spout. For example, the
> MapReduce(m,r,X) operation will consist of 2 bolts: a bolt with a map
> function m (a Bolt subclass with an attribute m) that applies m to the input
> data, and another similar bolt with the reduce function r. These two bolts
> will be connected in the Storm topology with a Storm group-by operation. The
> query evaluator will first generate a Storm topology from a physical plan and
> then will use the Storm evaluator to evaluate the topology. The only new
> components needed for these extensions are: 1) Bolt and Spout subclasses to
> capture MRQL operators 2) a Storm topology generator from MRQL plans.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
