[
https://issues.apache.org/jira/browse/MAPREDUCE-1126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12805779#action_12805779
]
Aaron Kimball commented on MAPREDUCE-1126:
------------------------------------------
Why are proposals now focusing on allowing users to specify different
serialization *factories?*
If allowing users to specify the use of a particular {{SerializationBase}} via
a flexible metadata map is considered too obscure, then I feel like the notion
of having separate {{SerializationFactory}} instances seems to be an
unnecessary level of abstraction. The current {{SerializationFactory}}
implemented in hadoop-common allows access to all {{SerializationBase}}
instances. If the focus is on user-accessibility of API, asking users to define
a SerializationFactory which will only produce a single {{SerializationBase}}
(e.g., {{WritableSerialization}} or {{AvroGenericSerialization}}) requires
needless one-off code, and clutters the class hierarchy.
Instead, I might understand adding an API such as
{{Job.setSerializationBase(ctxt, SerializationBase}}) where users directly set
the {{SerializationBase}} instance to use in a given context, and disregard the
{{SerializationFactory}} entirely.
For what it's worth, the patches that I and Tom have produced all make use of
the default {{SerializationFactory}} in Hadoop; this API then uses the metadata
map as defined in HADOOP-6165 to acquire the user's desired
{{SerializationBase}} instance as appropriate for each of map output key,
value, etc.
> shuffle should use serialization to get comparator
> --------------------------------------------------
>
> Key: MAPREDUCE-1126
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-1126
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: task
> Reporter: Doug Cutting
> Assignee: Aaron Kimball
> Fix For: 0.22.0
>
> Attachments: MAPREDUCE-1126.2.patch, MAPREDUCE-1126.3.patch,
> MAPREDUCE-1126.4.patch, MAPREDUCE-1126.5.patch, MAPREDUCE-1126.6.patch,
> MAPREDUCE-1126.patch, MAPREDUCE-1126.patch
>
>
> Currently the key comparator is defined as a Java class. Instead we should
> use the Serialization API to create key comparators. This would permit,
> e.g., Avro-based comparators to be used, permitting efficient sorting of
> complex data types without having to write a RawComparator in Java.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
