Transport client / Two hop

Elastic Stack Elasticsearch
1

Hi,

I test two configurations:

  1. C1: An index with 1 primary shard and 0 replica shard. Primary shard
    is on one server. The transport client is configured with the ES instance.
  2. C2: An index with 1 primary shard and 1 replica shard. Primary shard
    is on one server and replica on another. The transport client is configured
    with the 2 ES instances.

Network is monitored during a search load:

  1. C1 : Reception ≈ 5Mb/s | Transmission ≈ 180Mb/s
  2. C2 : For each server : Reception ≈ 50Mb/s | Transmission ≈ 180Mb/s

I'm wondering why such a difference between C1 and C2 reception? Is what
the "two hop" means? Does it can explain that I get only a 25% request/s
boost between C1 and C2? Is the node client can improve this?

Thanks for your help!

Nicolas

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

2

Here is a gatling results of the C2 configuration :

At 13:02:50 I stop the replica shard. There is small difference of the
number of query/s. At 13:03:25 I restart the replica shard.

Le mercredi 29 mai 2013 10:24:50 UTC+2, Nicolas Labrot a écrit :

Hi,

I test two configurations:

  1. C1: An index with 1 primary shard and 0 replica shard. Primary
    shard is on one server. The transport client is configured with the ES
    instance.
  2. C2: An index with 1 primary shard and 1 replica shard. Primary
    shard is on one server and replica on another. The transport client is
    configured with the 2 ES instances.

Network is monitored during a search load:

  1. C1 : Reception ≈ 5Mb/s | Transmission ≈ 180Mb/s
  2. C2 : For each server : Reception ≈ 50Mb/s | Transmission ≈ 180Mb/s

I'm wondering why such a difference between C1 and C2 reception? Is what
the "two hop" means? Does it can explain that I get only a 25% request/s
boost between C1 and C2? Is the node client can improve this?

Thanks for your help!

Nicolas

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.