Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
Hive >> mail # dev >> Review Request 12480: HIVE-4732 Reduce or eliminate the expensive Schema equals() check for AvroSerde


+
Mohammad Islam 2013-07-11, 22:31
Copy link to this message
-
Re: Review Request 12480: HIVE-4732 Reduce or eliminate the expensive Schema equals() check for AvroSerde

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/12480/#review23113
-----------------------------------------------------------
Do you have after-optimization performance numbers?  Can you add a test to verify that the reencoder cache is working correctly?  Feed in a record with one uuid, then another with a different and verify that the cache has two elements.  Adding a third record with the original UUID shouldn't increase the size of the cache.  Also, that adding n records all with the same schema creates only one reencoder...
serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroDeserializer.java
<https://reviews.apache.org/r/12480/#comment46953>

    verifiedRecordReaders -> noReencodingNeeded ?

serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroDeserializer.java
<https://reviews.apache.org/r/12480/#comment46956>

    readability: pull out getRecordReaderID into its own var

serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroGenericRecordWritable.java
<https://reviews.apache.org/r/12480/#comment46958>

    Need to write out the uuid too

serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroGenericRecordWritable.java
<https://reviews.apache.org/r/12480/#comment46959>

    Need to read in the uuid too
- Jakob Homan
On July 11, 2013, 3:31 p.m., Mohammad Islam wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/12480/
> -----------------------------------------------------------
>
> (Updated July 11, 2013, 3:31 p.m.)
>
>
> Review request for hive, Ashutosh Chauhan and Jakob Homan.
>
>
> Bugs: HIVE-4732
>     https://issues.apache.org/jira/browse/HIVE-4732
>
>
> Repository: hive-git
>
>
> Description
> -------
>
> From our performance analysis, we found AvroSerde's schema.equals() call consumed a substantial amount ( nearly 40%) of time. This patch intends to minimize the number schema.equals() calls by pushing the check as late/fewer as possible.
>
> At first, we added a unique id for each record reader which is then included in every AvroGenericRecordWritable. Then, we introduce two new data structures (one hashset and one hashmap) to store intermediate data to avoid duplicates checkings. Hashset contains all the record readers' IDs that don't need any re-encoding. On the other hand, HashMap contains the already used re-encoders. It works as cache and allows re-encoders reuse. With this change, our test shows nearly 40% reduction in Avro record reading time.
>  
>    
>
>
> Diffs
> -----
>
>   ql/src/java/org/apache/hadoop/hive/ql/io/avro/AvroGenericRecordReader.java dbc999f
>   serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroDeserializer.java c85ef15
>   serde/src/java/org/apache/hadoop/hive/serde2/avro/AvroGenericRecordWritable.java 66f0348
>   serde/src/test/org/apache/hadoop/hive/serde2/avro/TestSchemaReEncoder.java 9af751b
>   serde/src/test/org/apache/hadoop/hive/serde2/avro/Utils.java 2b948eb
>
> Diff: https://reviews.apache.org/r/12480/diff/
>
>
> Testing
> -------
>
>
> Thanks,
>
> Mohammad Islam
>
>

+
Mohammad Islam 2013-07-15, 23:47
+
Mohammad Islam 2013-07-15, 23:48
+
Mohammad Islam 2013-08-07, 02:13
+
Jakob Homan 2013-08-26, 05:35
+
Mohammad Islam 2013-08-30, 18:52
+
Mohammad Islam 2013-08-30, 18:49
+
Jakob Homan 2013-09-12, 22:33