[
https://issues.apache.org/jira/browse/HBASE-13408?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15595991#comment-15595991
]
stack commented on HBASE-13408:
-------------------------------
[~ram_krish], [~anoopsamjohn], [~eshcar], [~ebortnik] and myself met last Weds
morning, the 19th of October to chat about where we are all at on inmemory
compaction. Here are some rough notes:
{code}
1. Reiteration that inmemory compaction needs to be on all the time with no
associated perf degradation and with minimal config required to get benefit.
2. Need more testing and with more variety (Zipfian so inmemory compaction gets
a chance to shine). We'll all pitch in here.
3. When to merge (Eshcar "It is just a question of when.."). Back and forth.
Concern that current default of merge on every flush is a default that will
make inmemory compaction look bad because it for sure generates loads of
garbage. Suggestion to go to other extreme where we merge only at flush-to-disk
letting the pipeline build up in memory.
4. We'll pick up CellChunkMap at the next meeting but meantime wukk revive a
rumored existing umbrella issue. CCM is where we'll get biggest bang for the
buck so excited to get this done. Need to solve the cell-too-big-issue still.
Maybe split its dev some between Y! and Intel/St.Ack... TBD.
5. Ram and Anoop fixed some GC issues in "HBASE-16608 Introducing the ability
to merge ImmutableSegments without copy-compaction or SQM usage" and will put
up a new version of patch w/ fixes. Generally agreed patch is close to commit.
The Y! crew are on holiday until next Tuesday.
Did the note to the dev list go up on state of inmemory compaction (after
Tuesday -- smile).
...
The back-and-forth about when/how to merge will probably continue into
HBase-16417; we have to keep looking for the sweet spot.
{code}
> 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: HBASE-13408-trunk-v01.patch,
> HBASE-13408-trunk-v02.patch, HBASE-13408-trunk-v03.patch,
> HBASE-13408-trunk-v04.patch, HBASE-13408-trunk-v05.patch,
> HBASE-13408-trunk-v06.patch, HBASE-13408-trunk-v07.patch,
> HBASE-13408-trunk-v08.patch, HBASE-13408-trunk-v09.patch,
> HBASE-13408-trunk-v10.patch,
> HBaseIn-MemoryMemstoreCompactionDesignDocument-ver02.pdf,
> HBaseIn-MemoryMemstoreCompactionDesignDocument-ver03.pdf,
> HBaseIn-MemoryMemstoreCompactionDesignDocument-ver04.pdf,
> HBaseIn-MemoryMemstoreCompactionDesignDocument.pdf,
> InMemoryMemstoreCompactionEvaluationResults.pdf,
> InMemoryMemstoreCompactionMasterEvaluationResults.pdf,
> InMemoryMemstoreCompactionScansEvaluationResults.pdf,
> StoreSegmentandStoreSegmentScannerClassHierarchies.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)
