[
https://issues.apache.org/jira/browse/HBASE-13510?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14537550#comment-14537550
]
ramkrishna.s.vasudevan commented on HBASE-13510:
------------------------------------------------
bq.What is benefit of this patch?
The first benefit of this patch is that we are getting rid of compare(Cell,
byte[]) that comes with ROW_COL bloom so that removing such comparisons will
help us remove this compare(Cell, byte[]) API in CellComparator.
The ByteBloomFilter is not getting as a bloomfilter type making us remove the
dependency the bloom was having so that all the bloom types were operating on
byte[] though the ROW_COL was operating on the key part of a KV formed from row
and column.
bq.bloomKeyKV = BloomFilterUtil.createBloomKeyValue
ROW_COL will deal with a KeyValue(ie. a Cell) based on row and column part of a
keyvalue. So if we create a Cell any way to get the key byte[] we need to
convert that into a Keyvalue, hence instead of going with Cell and doing a
typecast to KV directly creating a KeyValue in this case is better. This key
byte[] is directly going to be used in the hash formation.
bq.Returning byte array of key-only portion?
Yes. We have to do that.
bq.I asked this before too.. .how many itimes we creating new keys here?
This we are doing it every time to form the key part of the ROW_COL bloom key.
This is nothing new, just using a static method instead of an instance method
previously.
bq.We have to add this to Interface? 55 boolean contains(KeyValue kvKey,
ByteBuffer bloom);
Ya. Same as my above comment. Going with Cell we anyway need to have the key
byte[] so that the hash functions can use them. Having it as a Cell would make
us to convert to a keyvalue anyway.
bq..e. preserve Writable?
Not doing any change there. The CompoundBloomFilterWriter needs these to add
the bloom meta data as part of the store file.
> Refactor Bloom filters to make use of Cell Comparators in case of ROW_COL
> -------------------------------------------------------------------------
>
> Key: HBASE-13510
> URL: https://issues.apache.org/jira/browse/HBASE-13510
> Project: HBase
> Issue Type: Sub-task
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Fix For: 2.0.0
>
> Attachments: HBASE-13510_1.patch, HBASE-13510_2.patch
>
>
> In order to address the comments over in HBASE-10800 related to comparing
> Cell with a serialized KV's key we had some need for that in Bloom filters.
> After discussing with Anoop, we found that it may be possible to
> remove/modify some of the APIs in the BloomFilter interfaces and for doing
> that we can purge ByteBloomFilter.
> I read the code and found that ByteBloomFilter was getting used in V1 version
> only. Now as it is obsolete we can remove this code and move some of the
> static APIs in ByteBloomFilter to some other util class or bloom related
> classes which will help us in refactoring the code too.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
