[
https://issues.apache.org/jira/browse/HBASE-10191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13908002#comment-13908002
]
Matt Corgan commented on HBASE-10191:
-------------------------------------
I hate to continue the tangent, but I'd add that even the occasional compaction
that CMS triggers is dependent on how many objects need to be compacted. It's
because "random" access memory isn't as random anymore because there are
enormous speed boosts when copying long swaths of sequential memory. So
compacting 100 1GB slabs should be far faster than compacting 1 billion 100B
KeyValues that are scattered around the heap. I also wonder if there's a slab
size big enough that hotspot won't bother moving it during a compaction (but i
have no idea).
Separately, one of the reasons Nick and I thought ByteRange should be an
interface was that we could back it with varying implementations including
arrays, HeapByteBuffers, DirectByteBuffers, netty ByteBufs, etc. A utility
similar to IOUtils.copy could help optimizing the copies between the different
implementations. Another advantage of using it as the primary interface is
that its internal compareTo method uses hbase-friendly unsigned byte
comparison, making it easy to put ByteRanges into traditional sorted
collections like TreeSet/CSLM without passing an external comparator.
I could see using an allocator based on huge on or off-heap slabs where smaller
pages/blocks are referenced by reusable ByteRanges. The allocator could
recycle memory by continuously picking the least utilized slab and copying
(moving) its occupied ByteRanges to the slab at the head of the queue. This
would provide constant compaction via fast sequential copying.
> Move large arena storage off heap
> ---------------------------------
>
> Key: HBASE-10191
> URL: https://issues.apache.org/jira/browse/HBASE-10191
> Project: HBase
> Issue Type: Umbrella
> Reporter: Andrew Purtell
>
> Even with the improved G1 GC in Java 7, Java processes that want to address
> large regions of memory while also providing low high-percentile latencies
> continue to be challenged. Fundamentally, a Java server process that has high
> data throughput and also tight latency SLAs will be stymied by the fact that
> the JVM does not provide a fully concurrent collector. There is simply not
> enough throughput to copy data during GC under safepoint (all application
> threads suspended) within available time bounds. This is increasingly an
> issue for HBase users operating under dual pressures: 1. tight response SLAs,
> 2. the increasing amount of RAM available in "commodity" server
> configurations, because GC load is roughly proportional to heap size.
> We can address this using parallel strategies. We should talk with the Java
> platform developer community about the possibility of a fully concurrent
> collector appearing in OpenJDK somehow. Set aside the question of if this is
> too little too late, if one becomes available the benefit will be immediate
> though subject to qualification for production, and transparent in terms of
> code changes. However in the meantime we need an answer for Java versions
> already in production. This requires we move the large arena allocations off
> heap, those being the blockcache and memstore. On other JIRAs recently there
> has been related discussion about combining the blockcache and memstore
> (HBASE-9399) and on flushing memstore into blockcache (HBASE-5311), which is
> related work. We should build off heap allocation for memstore and
> blockcache, perhaps a unified pool for both, and plumb through zero copy
> direct access to these allocations (via direct buffers) through the read and
> write I/O paths. This may require the construction of classes that provide
> object views over data contained within direct buffers. This is something
> else we could talk with the Java platform developer community about - it
> could be possible to provide language level object views over off heap
> memory, on heap objects could hold references to objects backed by off heap
> memory but not vice versa, maybe facilitated by new intrinsics in Unsafe.
> Again we need an answer for today also. We should investigate what existing
> libraries may be available in this regard. Key will be avoiding
> marshalling/unmarshalling costs. At most we should be copying primitives out
> of the direct buffers to register or stack locations until finally copying
> data to construct protobuf Messages. A related issue there is HBASE-9794,
> which proposes scatter-gather access to KeyValues when constructing RPC
> messages. We should see how far we can get with that and also zero copy
> construction of protobuf Messages backed by direct buffer allocations. Some
> amount of native code may be required.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
