You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@ignite.apache.org by siva <si...@bizruntime.com> on 2018/07/25 17:23:51 UTC

Partitioned vs Replicated mode

Hi,

We need to test the Partitioned vs Replicated cache modes.Since as per the
documentation 

Replicated use for small data sets and where read operations
frequently,Partitioned mode use for large data sets and where updates are
frequently.

But in our scenario each message we end up with more than 15kb ,and read and
update  operations frequently


Queries:
--------
1.What is the best way and framework  to test Partitioned vs Replicated mode
with large data sets?
2.I have tested with the 4 nodes in two machines, its giving almost same
results in both partitioned and replicated mode.(both Put and Get).I have
used dotTrace framework to test.Is there any other framework test?
2.Initially if we start the cluster with 3 server nodes, and created a cache
name called "Users" with one backup copy .Let say if later another 2  nodes
added ,can we increase the backup copies with out destroying that cache?




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Partitioned vs Replicated mode

Posted by Evgenii Zhuravlev <e....@gmail.com>.

Hi,

Looks like this thread can help you:
http://apache-ignite-users.70518.x6.nabble.com/Yardstick-benchmarking-on-Ignite-with-Persistent-store-td18963.html

Evgenii

2018-07-31 15:14 GMT+03:00 siva <si...@bizruntime.com>:

> Thank You for your reply,
>
> But, How to test with native parsistence?
>
> I have set a property by enabling native parsistnce in
> *ignite-base-config.xml*,when i start benchmark-run-all.sh ,its terminating
> after some time .
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>

Re: Partitioned vs Replicated mode

Posted by siva <si...@bizruntime.com>.

Thank You for your reply,

But, How to test with native parsistence?

I have set a property by enabling native parsistnce in
*ignite-base-config.xml*,when i start benchmark-run-all.sh ,its terminating
after some time .



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Partitioned vs Replicated mode

Posted by Evgenii Zhuravlev <e....@gmail.com>.

Hi,

1,2. You can use yardstick for benchmarking:
https://apacheignite.readme.io/docs/perfomance-benchmarking. It contains a
lot of already written benchmarks or you can write your own.
3. No, as for now, it's not possible to change backups count after the
cache was created.

Evgenii

2018-07-25 20:23 GMT+03:00 siva <si...@bizruntime.com>:

> Hi,
>
> We need to test the Partitioned vs Replicated cache modes.Since as per the
> documentation
>
> Replicated use for small data sets and where read operations
> frequently,Partitioned mode use for large data sets and where updates are
> frequently.
>
> But in our scenario each message we end up with more than 15kb ,and read
> and
> update  operations frequently
>
>
> Queries:
> --------
> 1.What is the best way and framework  to test Partitioned vs Replicated
> mode
> with large data sets?
> 2.I have tested with the 4 nodes in two machines, its giving almost same
> results in both partitioned and replicated mode.(both Put and Get).I have
> used dotTrace framework to test.Is there any other framework test?
> 2.Initially if we start the cluster with 3 server nodes, and created a
> cache
> name called "Users" with one backup copy .Let say if later another 2  nodes
> added ,can we increase the backup copies with out destroying that cache?
>
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>