[
https://issues.apache.org/jira/browse/HADOOP-3856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12619911#action_12619911
]
Raghu Angadi commented on HADOOP-3856:
--------------------------------------
Thanks Ankur.
Untill yesterday I was thinking of custom implementation, say leaner MINA :). I
actually started writing code. But it might just be better and faster to take
the parts we want from MINA make necessary changes. will talk to Owen today
about this approach.
Some important changes :
* Threading model : MINA's division of work is from pre-devpoll era I think. It
divides the connections rather than individual pieces work among the threads
since that divides the polling cost as well. But polling cost is virtually
zero. Also since some of our handlers do disk I/O, it is better not to tie one
connection to one worker thread. In DataNode's case we want to have a set of
threads per disk.
* We need another layer below IoSession. Mina's handlers deal with data already
read or written by Mina's core. We will have another layer in between where
another type of handlers can handle events like 'readReady()', writeReady(),
acceptReady() etc.
* Users can register existing sockets of course.
If this sounds good I will prepare an initial version and we can go from
there. I didn't look at xSockets since I didn't want it to influence my
implementation (I am not sure of GPL implications).
> Asynchronous IO Handling in Hadoop and HDFS
> -------------------------------------------
>
> Key: HADOOP-3856
> URL: https://issues.apache.org/jira/browse/HADOOP-3856
> Project: Hadoop Core
> Issue Type: New Feature
> Components: dfs, io
> Reporter: Raghu Angadi
> Assignee: Raghu Angadi
>
> I think Hadoop needs utilities or framework to make it simpler to deal with
> generic asynchronous IO in Hadoop.
> Example use case :
> Its been a long standing problem that DataNode takes too many threads for
> data transfers. Each write operation takes up 2 threads at each of the
> datanodes and each read operation takes one irrespective of how much activity
> is on the sockets. The kinds of load that HDFS serves has been expanding
> quite fast and HDFS should handle these varied loads better. If there is a
> framework for non-blocking IO, read and write pipeline state machines could
> be implemented with async events on a fixed number of threads.
> A generic utility is better since it could be used in other places like
> DFSClient. DFSClient currently creates 2 extra threads for each file it has
> open for writing.
> Initially I started writing a primitive "selector", then tried to see if such
> facility already exists. [Apache MINA|http://mina.apache.org] seemed to do
> exactly this. My impression after looking the the interface and examples is
> that it does not give kind control we might prefer or need. First use case I
> was thinking of implementing using MINA was to replace "response handlers" in
> DataNode. The response handlers are simpler since they don't involve disk
> I/O. I [asked on MINA user
> list|http://www.nabble.com/Async-events-with-existing-NIO-sockets.-td18640767.html],
> but looks like it can not be done, I think mainly because the sockets are
> already created.
> Essentially what I have in mind is similar to MINA, except that read and
> write of the sockets is done by the event handlers. The lowest layer
> essentially invokes selectors, invokes event handlers on single or on
> multiple threads. Each event handler is is expected to do some non-blocking
> work. We would of course have utility handler implementations that do read,
> write, accept etc, that are useful for simple processing.
> Sam Pullara mentioned that [xSockets|http://xsocket.sourceforge.net/] is more
> flexible. It is under GPL.
> Are there other such implementations we should look at?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
