Folks,
I worked on issue https://issues.apache.org/jira/browse/IGNITE-3018 that
is related to performance of Rendezvous AF.
But Wang/Jenkins hash integer hash distribution is worse then MD5. So, i
try to use simple partition balancer close
to Fair AF for Rendezvous AF.
Take a look at the heatmaps of distributions at the issue. e.g.:
- Compare of current Rendezvous AF and new Rendezvous AF based of
Wang/Jenkins hash:
https://issues.apache.org/jira/secure/attachment/12858701/004.png
- Compare of current Rendezvous AF and new Rendezvous AF based of
Wang/Jenkins hash with partition balancer:
https://issues.apache.org/jira/secure/attachment/12858690/balanced.004.png
When the balancer is enabled the distribution of partitions by nodes
looks like close to even distribution
but in this case there is not guarantee that a partition doesn't move
from one node to another
when node leave topology.
It is not guarantee but we try to minimize it because sorted array of
nodes is used (like in for pure-Rendezvous AF).
I think we can use new fast Rendezvous AF and use 'useBalancer' flag
instead of Fair AF.
On 03.03.2017 1:56, Denis Magda wrote:
What??? Unbelievable. It sounds like a design flaw to me. Any ideas
how to fix?
—
Denis
On Mar 2, 2017, at 2:43 PM, Valentin Kulichenko
<[email protected]
<mailto:[email protected]>> wrote:
Adding back the dev list.
Folks,
Are there any opinions on the problem discussed here? Do we really
need FairAffinityFunction if it can't guarantee cross-cache collocation?
-Val
On Thu, Mar 2, 2017 at 2:41 PM, vkulichenko
<[email protected]
<mailto:[email protected]>> wrote:
Hi Alex,
I see your point. Can you please outline its advantages vs rendezvous
function?
In my view issue discussed here makes it pretty much useless in vast
majority of use cases, and very error-prone in all others.
-Val
--
View this message in context:
http://apache-ignite-users.70518.x6.nabble.com/Same-Affinity-For-Same-Key-On-All-Caches-tp10829p11006.html
<http://apache-ignite-users.70518.x6.nabble.com/Same-Affinity-For-Same-Key-On-All-Caches-tp10829p11006.html>
Sent from the Apache Ignite Users mailing list archive at
Nabble.com <http://Nabble.com>.
--
Taras Ledkov
Mail-To: [email protected]
