[infinispan-dev] Denormalizing hashes

Wed Dec 11 04:01:23 EST 2013

On Wed, Dec 11, 2013 at 10:38 AM, Radim Vansa <rvansa at redhat.com> wrote:

>  Hi Dan
>
> I am not speaking about changing something for the C++ client, I
> understand that the client code cannot be changed in order to keep the
> backward compatibility.
>

Sure, I was just trying to give some background information on what we
discussed and why we still have the wheel-based CH in the client.

>
> The current hash-wheel approach is working well, but there are few flaws
> that could be fixed keeping the client code untouched. Please, correct me
> if I am wrong.
>
> 1) The denormalization is executed for every client for every topology
> change/client join. I don't have any numbers, but calling the hashing
> algorithm million times per every such occasion sounds as wasting computing
> power. -> cache the denormalized stuff on server
>

+1, like I said it would be easy to do but it never came up as a problem
before.

>
> 2) The server is sending numOwners hashIds per segment, one for each
> owner. What's the reason for that? I think that only primary owners should
> be inserted there. This would:
>

The main reason is to support clients from Infinispan 5.1, which pick a
random owner instead of always choosing the primary (
https://issues.jboss.org/browse/ISPN-2655).

> a) target all PUT requests to primary owner, reducing PUT latency and
> lowering the general load in cluster
>

Nope, it wouldn't. The same fraction of requests would go to the primary
owner as before, because we won't find the exact "denormalized" hash id
that maps to the segment border when normalized.

b) reduce the routing information
>

For 7.0, I guess we could say that 5.1 clients are no longer supported and
we could switch to sending only the primary owners to the clients. But I'm
not sure whether the loss of backwards compatibility is worth a couple
hundred bytes sent once for every client.

> And yes, ISPN-3530 and ISPN-3701 are pretty serious, but IMO rather
> orthogonal to the segment vs. hash wheel approach and its details.
>
>
Agree. Could you create issues in JIRA for both your proposals?

>  Radim
>
>
>
> On 12/11/2013 09:18 AM, Dan Berindei wrote:
>
>  Hi Radim
>
>  Actually, it's me that wrote the denormalization code :)
>
>  It was meant as a stop-gap measure before we upgraded the HotRod protocol
> to support the segment-based consistent hash, but the denormalization
> worked well enough (or so we thought) that we didn't get to changing the
> protocol yet.
>
> That's not a big change in itself, but we also wanted to make the
> consistent hash per-cache on the client (it's now per-cache manager), and
> that's a bit more complicated to do. And it's not like it would have been a
> good idea to change this before starting the C++ client, the client would
> still have to support the current style of consistent hash.
>
>
> On Tue, Dec 10, 2013 at 8:17 PM, Radim Vansa <rvansa at redhat.com> wrote:
>
>> Hi Galder,
>>
>> as I am trying to debug some problem in C++ client, I was looking into
>> the server code. And I am not sure whether I understand the code
>> correctly, but it seems to me that the server denormalizes the
>> consistent hash for each client anew (after each topology change or
>> client joining). Is this true? Looking into trace logs, I can see stuff
>> like
>>
>> 18:15:17,339 TRACE [org.infinispan.server.hotrod.Encoders$Encoder12$]
>> (HotRodServerWorker-12) Writing hash id 639767 for 192.168.11.101:11222
>>
>>  From denormalizeSegmentHashIds() method I see that this means that we
>> have executed the hash function 639768 times just to notify one client.
>> Is my understanding correct?
>>
>
>  Yes, this happens every time a client joins and/or every time the cache
> topology changes.
>
> We could easily cache the result of denormalizeSegmentHashIds, as it only
> depends on the number of segments. It's just that I wasn't expecting it to
> take so many iterations.
>
>
>
>>
>> Also, there is nothing like the concept of primary owner, is this right?
>>
>
>  The client CH doesn't have a concept of backup owners. But for each
> (hash id, server) pair that gets sent to the client, it means all the hash
> codes between the previous hash id and this hash id have this server as the
> primary owner. The server in the next (hash id, server) pair is the first
> backup, and so on.
>
>  For each segment, the server generates numOwners (hash id, server)
> pairs. That means, for most of the hash codes in the segment, the list of
> owners on the client will be the same as the list of owners on the server.
> But for 0.0002 (leewayFraction) of the hash codes, the client primary owner
> will be indeed one of the server backup owners.
>
>
>
>> I thought that every first request in HotRod will go to primary owner,
>> so that the PUT does not have to do the first hop and is executed
>> directly on the primary. But it seems to me that it goes to any of the
>> owners (practically random one, as you are only looking for the numOwner
>> ids in leeway = on the beginning of the range - then, 99.98% or more
>> requests should go to the server with last position in the leeway). This
>> looks pretty suboptimal for writes, isn't it?
>>
>
>  I'm not sure what you mean here, but I'm pretty sure the request goes to
> the correct server because we have a test for it:
> ConsistentHashV1IntegrationTest
>
>  Cheers
>  Dan
>
>
>
>>
>> Cheers
>>
>> Radim
>>
>> PS: for every line of code you write in Scala, God kills a kitten
>>
>> --
>> Radim Vansa <rvansa at redhat.com>
>> JBoss DataGrid QA
>>
>> _______________________________________________
>> infinispan-dev mailing list
>> infinispan-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>>
>
>
>
> _______________________________________________
> infinispan-dev mailing listinfinispan-dev at lists.jboss.orghttps://lists.jboss.org/mailman/listinfo/infinispan-dev
>
>
>
> --
> Radim Vansa <rvansa at redhat.com> <rvansa at redhat.com>
> JBoss DataGrid QA
>
>
> _______________________________________________
> infinispan-dev mailing list
> infinispan-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/infinispan-dev/attachments/20131211/a115deb8/attachment-0001.html 