Elasticsearch Write performance for full text record

Hi All,

I am indexing text records into Elasticsearch, the size of each record are
20KB. We are getting a speed of 400 records second.

I have 3 machines ES cluster (32 GB RAM, 16 core, 2 disks) and 2 ES nodes
are running on each machine.

The minimum write speed required for our usecase is 2000 to 3000 records
per second.

Please share some tuning parameters that can increase the write performance.

Thanks,
Ankit Jain

--
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.

Can you describe your system in more detail before getting into "tuning"?

System:

How fast are your disks?
Do you connect from remote machine over the network? How fast is the
network connection?
What Java JVM version do you use and how did you configure it?
How many documents do you want to index?

ES:

What ES client do you use?
How did you configure the settings for bulk indexing?
How many threads do you use in parallel?
What machines do you connect to when you index?

And:
Why do you run two nodes on a single machine? (The answer is simple, it
makes no sense)

Jörg

Am 29.04.13 18:54, schrieb Ankit Jain:

Hi All,

I am indexing text records into Elasticsearch, the size of each record
are 20KB. We are getting a speed of 400 records second.

I have 3 machines ES cluster (32 GB RAM, 16 core, 2 disks) and 2 ES
nodes are running on each machine.

The minimum write speed required for our usecase is 2000 to 3000
records per second.

Please share some tuning parameters that can increase the write
performance.

Thanks,
Ankit Jain

--
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.

Hi Jorg,

Thanks for the reply..

Do you connect from remote machine over the network? How fast is the
network connection?
Using Gbps network

What Java JVM version do you use and how did you configure it?
Java1.6.0_24

How many documents do you want to index?
We want to index 10 millions documents in 1 index(Total indices=24)

What ES client do you use?
We are using ES transport client.

How did you configure the settings for bulk indexing?
BulkRequestBuilder bulkRequest = client.prepareBulk();

bulkRequest.add(client.prepareIndex("twitter", "tweet", "1")
.setSource(jsonBuilder()
.startObject()
.field("user", "kimchy")
.field("postDate", new Date())
.field("message", "trying out Elastic Search")
.endObject()
)
);

How many threads do you use in parallel?
We are using 8 threads: Our application are running on 2 machines (4
threads on each machine)

How fast your disks?
we are using SATA disk.

Please suggest some tuning parameters....

Thanks,
Ankit Jain

On Monday, 29 April 2013 22:38:09 UTC+5:30, Jörg Prante wrote:

Can you describe your system in more detail before getting into "tuning"?

System:

How fast are your disks?
Do you connect from remote machine over the network? How fast is the
network connection?
What Java JVM version do you use and how did you configure it?
How many documents do you want to index?

ES:

What ES client do you use?
How did you configure the settings for bulk indexing?
How many threads do you use in parallel?
What machines do you connect to when you index?

And:
Why do you run two nodes on a single machine? (The answer is simple, it
makes no sense)

Jörg

Am 29.04.13 18:54, schrieb Ankit Jain:

Hi All,

I am indexing text records into Elasticsearch, the size of each record
are 20KB. We are getting a speed of 400 records second.

I have 3 machines ES cluster (32 GB RAM, 16 core, 2 disks) and 2 ES
nodes are running on each machine.

The minimum write speed required for our usecase is 2000 to 3000
records per second.

Please share some tuning parameters that can increase the write
performance.

Thanks,
Ankit Jain

--
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.