[
https://issues.apache.org/jira/browse/HDFS-15553?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17188628#comment-17188628
]
Wang, Xinglong commented on HDFS-15553:
---------------------------------------
[~kihwal] Thank you for the comment. It is really helpful.
{quote}Also there are write calls that initially acquire the read lock then
reacquire the write lock.
{quote}
Yes, I do notice this, the addBlock rpc is a such example to get read lock and
then write lock. It's a trouble maker. Currently, I think as for the read lock
part, it can share with other read rpc. So I put it in my read rpc queue.
For the internal rpc calls, I agree with you. E.g, IRB may have a impact on
complete rpc call result if there is a reorder. For other scenarios, I don't
have concrete case for now.
{quote} are you using async edit logging and audit logging?
{quote}
Yes. We are using both. And we have some enhancement for async edit logging
part. I can share in another thread.
{quote} -HDFS-9198-
{quote}
This patch is also in our production env. As for the 1024 queue size, I don't
have data right now. Let me grab some logs and get back.
For Lock combination, it can't be too small neither can be too big. This is
tricky part. :)
{quote}Key to the success of this approach would depend on how smart the
dynamic read/write allocation mechanism works. This may be less critical if the
workload pattern is easily predictable or slowly changing, or if you want to
enforce a certain ratio or priority between reads and writes. In environments
where the workload is highly varied, there might be difficulty utilizing this
in its fullest extent.
{quote}
Yes. I can share my version of the dequeue strategy later by uploading some
code as well as some benchmark. It's very like FairCallQueue strategy a weight
based rpc scheduler. [~hexiaoqiao]
One of our cluster is mostly with 10:1 read to write ratio traffic. And this
approach will indeed not bring much benefit for cases where there are too many
read or there are too many write. It's designed for moderate case.
> Improve NameNode RPC throughput with ReadWriteRpcCallQueue
> -----------------------------------------------------------
>
> Key: HDFS-15553
> URL: https://issues.apache.org/jira/browse/HDFS-15553
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Reporter: Wang, Xinglong
> Priority: Major
>
> *Current*
> In our production cluster, a typical traffic model is read to write raito is
> 10:1 and sometimes the ratios goes to 30:1.
> NameNode is using ReEntrantReadWriteLock under the hood of FSNamesystemLock.
> Read lock is shared lock while write lock is exclusive lock.
> Read RPC and Write RPC comes randomly to namenode. This makes read and write
> mixed up. And then only a small fraction of read can really share their read
> lock.
> Currently we have default callqueue and faircallqueue. And we can
> refreshCallQueue on the fly. This opens room to design new call queue.
> *Idea*
> If we reorder the rpc call in callqueue to group read rpc together and write
> rpc together, we will have sort of control to let a batch of read rpc come to
> handlers together and possibly share the same read lock. Thus we can reduce
> Fragments of read locks.
> This will only improve the chance to share the read lock among the batch of
> read rpc due to there are some namenode internal write lock is out of call
> queue.
> Under ReEntrantReadWriteLock, there is a queue to manage threads asking for
> locks. We can give an example.
> R: stands for read rpc
> W: stands for write rpc
> e.g
> RRRRWRRRRWRRRRWRRRRWRRRRWRRRRWRRRRWRRRRW
> In this case, we need 16 lock timeslice.
> optimized
> RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRWWWWWWWW
> In this case, we only need 9 lock timeslice.
> *Correctness*
> Since the execution order of any 2 concurrent or queued rpc in namenode is
> not guaranteed. We can reorder the rpc in callqueue into read group and write
> group. And then dequeue from these 2 queues by a designed strategy. let's say
> dequeue 100 read and then dequeue 5 write rpc and then dequeue read again and
> then write again.
> Since FairCallQueue also does rpc call reorder in callqueue, for this part I
> think they share the same logic to guarantee rpc result correctness.
> *Performance*
> In test environment, we can see a 15% - 20% NameNode RPC throughput
> improvement comparing with default callqueue.
> Test traffic is 30 read:3 write :1 list using NNLoadGeneratorMR
> This performance is not a surprise. Due to some write rpc is not managed in
> callqueue. We can't do reorder to them by reording calls in callqueue.
> But still we can do a fully read write reorder if we redesign
> ReEntrantReadWriteLock to achieve this. This will be further step after this.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
