[
https://issues.apache.org/jira/browse/HBASE-10191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13852436#comment-13852436
]
Matt Corgan commented on HBASE-10191:
-------------------------------------
Something to keep in mind is that GC pauses can be influenced as much or more
by the number of live objects as they can by the raw size of the heap. 32GB of
block cache could be made of only 1mm 32KB blocks. This particular 32GB of
memory may not stop the world for very long. It's all the small remaining
objects that are keeping the garbage collector busy, and I bet the biggest
culprit here is the individual KeyValues in the memstores.
MemstoreLAB combines the backing arrays into big chunks to reduce heap
fragmentation, but there is still one object per KeyValue, and each object
needs to be considered by the collector. A big heap has big memstores, which
have lots of KeyValues - possibly far more than the 1mm blocks in the block
cache. A big advantage of flattening the memstores into blocks of key values
is that you might be reducing ~500 KeyValues to a single block object. This
500x reduction in objects strikes me as a significant GC pause improvement that
is independent from off-heap techniques.
----
Moving blocks off-heap and operating on them directly will be very cool.
DataBlockEncoders should be able to read off-heap blocks similarly to how they
do now, namely, copying only the modified bytes from the previous cell into an
array buffer. Vladimir makes a good point that it would be tough to match the
scan performance of unencoded data, so that would need some thinking.
> 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
(v6.1.4#6159)
