Resurrecting this topic: I've put some sample code on my fork here: https://github.com/matthoffman/infinispan/tree/dfjThere's a README there that offers an overview. I hope to do some performance comparisons over the weekend, but in the meantime, the code is there if you're curious.I'm mainly interested whether this is the type of thing that Infinispan may be interested in, as an alternative to the current distributed executor (much like the ForkJoinPool in the JDK is an alternative to a traditional thread pool).If so, the next steps would be to do some performance tests, pick one implementation to move forward with (out of the 3 prototypes on that branch) and clean it up sufficiently to be considered for a pull request.If its not something that Infinispan is interested in, then I'll change approaches and make it more generic, so that it isn't Infinispan-specific and can be used with other transports. I'm open to either option... I think there are pros and cons either way.Thanks!matt_______________________________________________On Mon, Mar 4, 2013 at 7:07 AM, Paolo Romano <romano@inesc-id.pt> wrote:
This sounds really interesting Matt. In the Cloud-TM project (www.cloudtm.eu) we are currently developing a parallel graph-analysis algorithm on top of Infinispan's DEF. I would be really curious to take a look at the framework you developed, and see how it could be exploited in our application.
Regards,
Paolo
-- Paolo Romano, PhD Coordinator of the Cloud-TM ICT FP7 Project (www.cloudtm.eu) Senior Researcher @ INESC-ID (www.inesc-id.pt) Assistant Professor @ Instituto Superior Tecnico (www.ist.utl.pt) Rua Alves Redol, 9 1000-059, Lisbon Portugal Tel. + 351 21 3100300 Fax + 351 21 3145843 Webpage http://www.gsd.inesc-id.pt/~romanop
On 3/2/13 5:40 PM, matt hoffman wrote:
Hey guys,
I've been working on a prototype of integrating Infinispan into our app. We do a lot of distributed processing across a small cluster, so I've played with Infinispan's existing distributed execution framework (which is nice), as well as using Infinispan alongside a normal message queue to distribute tasks. But I've also put together a prototype of a new distributed execution framework using fork-join pools that you all might be interested in. If it sounds like something that would be worthwhile for Infinispan, I can raise a Jira and submit a pull request with what I have so far. I'd need to get the CA and company policy stuff finalized; that might take a couple days. Meanwhile, in case there is any interest, I've described the approach I've taken below.
First, a little background:
A while back I worked on a side project that integrated a distributed work-stealing algorithm into the standard JDK fork-join queue. It used JGroups for communication, because it was quick and easy for prototyping. So this week I thought i'd take a stab at porting that over to Infinispan. The algorithm I came up with for Infinispan is a bit less of a work-stealing algorithm, to take advantage of Infinispan's built-in distribution capabilities, but I think it's still fairly efficient.
My basic approach was to take in a cache in the constructor, much like the existing distributed executor, and then create a parallel, DIST-mode cache that uses the same hash & grouping configuration as the original cache. That new parallel cache is the "task cache", and we use that to distribute available tasks across the cluster. It's a distributed cache so that tasks are partitioned across a large cluster, and it uses the hashing config of the original cache and a KeyAffinityService to attempt to distribute the tasks to the same nodes that contain the data being worked on. Nodes use cache listeners to be notified when there is new work available, and the atomic replace() to "check out" the tasks for execution, and "check in" the results.
The basic algorithm is something like this:
For a refresher, a normal FJ pool has a fork() method that takes in a task, and then places that task on an internal queue (actually, one of several queues). When threads are idle, they look to the nearest work queue for work. If that work queue does not have work, they "steal" work from another thread's queue. So in the best case, tasks remain on the same thread as the task that spawned them, so tasks that process the same data as their parents may still have that data in the CPU's cache, etc. There's more to it than that, but that's the basic idea.
This distributed algorithm just adds an extra layer on top for tasks that are marked "distributable" (by extending DistributedFJTask instead of the normal ForkJoinTask). When you call fork() with a DistributedFJTask, it first checks to see if the local pool's work queue is empty. If so, we just go ahead and submit it locally; there's no reason to distribute it. If not, we put the task in the task cache, and let Infinispan distribute it. When a node has no more work to do in its internal fork-join queues, it looks at the task cache and tries to pull work from there.
So, it isn't really a "work-stealing" algorithm, per se; the distributable tasks are being distributed eagerly using Infinispan's normal cache distribution. But I'm hoping that doing that also makes it easier to handle node failure, since nodes collectively share a common picture of the work to be done.
This approach required one change to the actual FJ classes themselves (in org.infinispan.util.concurrent.jdk8backported). That's probably the most controversial change. I had to make the original ForkJoinTask's fork() method non-final in order to extend it cleanly. There's probably a way around that, but that's the cleanest option I have thought of thus far.
And lastly, it's not done yet: basic task distribution is working, but I haven't tackled failover to any real extent yet. The biggest questions, though, are around what to do with the existing distributed execution interfaces. For example, DistributedTask has a getCallable() method because it assumes it's wrapping a Callable. But ForkJoinTasks don't extend Callable. I could put in a shim to wrap the DistributedFJTasks into Callables for the sake of that method, but I don't know if it's worth it. Similarly, the DistributedExecutorService interface exposes a lot of submit-to-specific-address or submit-to-all-addresses methods, which are an odd fit here since tasks are distributed via their own cache. Even if I used a KeyAffinityService to target the task to the given Address, it might get picked up by another node that shares that same hash. But I can add in a direct-to-single-Address capability in if that seems worthwhile. Alternately, I can just use entirely different interfaces (DistributedFJExecutorService, DistributedFJTask?).
Thoughts? Concerns? Glaring issues?
_______________________________________________ infinispan-dev mailing list infinispan-dev@lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev
_______________________________________________
infinispan-dev mailing list
infinispan-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev
infinispan-dev mailing list
infinispan-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev