Re: Cluster performance with data node down

Jon Stephens wrote:
>
> Try repeating your test using 4 data nodes (and NoOfReplicas=2) to
> start with rather than just 2. I think you'll find when you drop one
> DN that your performance won't decrease nearly so much.
>
> More data nodes -> greater parallelisation.
>
> Conversely, 1 data node -> *no* parallelisation.
>
> So what you're observing with 2 DN -> 1 DN isn't really "abnormal".
>
> cheers
>
> jon.

First of all thanks for the answer.

Secondly I must admit that what you say doesn't sound very convincing to me.

What we are talking about here is a query which on a single machine configuration (no cluster) takes less than .3 sec. to complete.

So we have
Time to complete select query :

*) single machine - no cluster : < 0.3 seconds
*) cluster - 2/2 data nodes : < 0.4 seconds
*) cluster - 1/2 data nodes : > 12 seconds.

What we have here is a configuration without any workload (except the query I execute). As far as I can tell there is no task here which can benefit from parallelization.

But even if it were, parallelization could in no way account for a difference in execution of a factor of 30 (12 / 0.4)

As a software architect my guess is that here we have something of an anomaly in the way things are managed in a degraded cluster.

Do you need help?

In addition to our free technical library, Pantek provides professional IT services to companies who utilize Open Source software: Learn more about Pantek's Technical Support offerings Place an order for 24/7/365 Technical Support here Learn more about Pantek, who we are, and what we do Chat with a Live Operator Related lists.mysql.com Links announce benchmarks bugs certification cluster commits dotnet gui-tools internals java maxdb myodbc mysql mysqldoc packagers perl plusplus replication win32 Request more information about Pantek

I could also add data nodes, but unless some of you explains to me how this could really set things right I'm still thinking we are barking at the wrong tree...

-- 
MySQL Cluster Mailing List
For list archives: 
http://lists.mysql.com/cluster
To unsubscribe:    
http://lists.mysql.com/cluster?unsub=lists@pantek.com