[
https://issues.apache.org/jira/browse/HBASE-13408?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14646101#comment-14646101
]
Duo Zhang commented on HBASE-13408:
-----------------------------------
Things we talking here are all 'In Memory', so I do not think we need to modify
WAL...
I think all the logic could be down in a special memstore implementation? For
example, you can set the flush-size to 128M, and introduce a compact-size which
only consider the active set size to 32M. When you find the active set reaches
32M then you put it into pipeline and try to compact segments in pipeline to
reduce memory usage. The upper layer does not care about how many segments you
have, it only cares about the total memstore size. If it reaches 128M then a
flush request is coming, then you should flush all data to disk. If there are
many redundant cells then the total memstore will never reaches 128M, I think
this is exactly what we want here? And this way you do not change the semantic
of flush, the log truncating should also work as well.
And I think you can use some more compact data structures instead of skip list
since the segments in pipeline are read only? This may bring some benefits even
if we do not have many redundant cells.
What do you think? [~eshcar]. Sorry a bit late. Thanks.
> HBase In-Memory Memstore Compaction
> -----------------------------------
>
> Key: HBASE-13408
> URL: https://issues.apache.org/jira/browse/HBASE-13408
> Project: HBase
> Issue Type: New Feature
> Reporter: Eshcar Hillel
> Attachments:
> HBaseIn-MemoryMemstoreCompactionDesignDocument-ver02.pdf,
> HBaseIn-MemoryMemstoreCompactionDesignDocument.pdf,
> InMemoryMemstoreCompactionEvaluationResults.pdf
>
>
> A store unit holds a column family in a region, where the memstore is its
> in-memory component. The memstore absorbs all updates to the store; from time
> to time these updates are flushed to a file on disk, where they are
> compacted. Unlike disk components, the memstore is not compacted until it is
> written to the filesystem and optionally to block-cache. This may result in
> underutilization of the memory due to duplicate entries per row, for example,
> when hot data is continuously updated.
> Generally, the faster the data is accumulated in memory, more flushes are
> triggered, the data sinks to disk more frequently, slowing down retrieval of
> data, even if very recent.
> In high-churn workloads, compacting the memstore can help maintain the data
> in memory, and thereby speed up data retrieval.
> We suggest a new compacted memstore with the following principles:
> 1. The data is kept in memory for as long as possible
> 2. Memstore data is either compacted or in process of being compacted
> 3. Allow a panic mode, which may interrupt an in-progress compaction and
> force a flush of part of the memstore.
> We suggest applying this optimization only to in-memory column families.
> A design document is attached.
> This feature was previously discussed in HBASE-5311.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
