Elasticsearch high availability

Elastic Stack Elasticsearch
1

Hi All,

We want to deploy Elasticsearch in production environment, we would like to
achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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

Hey,

I am sorry, but I dont understand you. Why do you refrain from using
replication, even though it can solve your problem of high availability? Do
you think replication is bad for you because of other reasons?
Maybe you can shed some light on this.

Thx!

--Alex

On Fri, May 17, 2013 at 11:16 AM, Ankit Jain ankitjaincs06@gmail.comwrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like
to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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.

--
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.

3

Replication is the key to high availability. Doesn't matter if it is a
search engine or a database or static content. You don't put all your eggs
in one basket.

--
Ivan

On Fri, May 17, 2013 at 3:12 AM, Alexander Reelsen alr@spinscale.de wrote:

Hey,

I am sorry, but I dont understand you. Why do you refrain from using
replication, even though it can solve your problem of high availability? Do
you think replication is bad for you because of other reasons?
Maybe you can shed some light on this.

Thx!

--Alex

On Fri, May 17, 2013 at 11:16 AM, Ankit Jain ankitjaincs06@gmail.comwrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like
to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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.

--
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.

--
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.

4

Hi Alexander,

We don''t want to use replication, because replication will increase the
storage cost.

Thanks,
Ankit Jain

On Friday, 17 May 2013 15:42:06 UTC+5:30, Alexander Reelsen wrote:

Hey,

I am sorry, but I dont understand you. Why do you refrain from using
replication, even though it can solve your problem of high availability? Do
you think replication is bad for you because of other reasons?
Maybe you can shed some light on this.

Thx!

--Alex

On Fri, May 17, 2013 at 11:16 AM, Ankit Jain <ankitj...@gmail.com<javascript:>

wrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like
to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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 elasticsearc...@googlegroups.com <javascript:>.
For more options, visit https://groups.google.com/groups/opt_out.

--
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.

5

So, you don't want to backup your data because of the hard disk cost?
So, you have to pay the cost of loosing your data when a node crash.

If the node stops and then restart, it will reuse this node data.
If you loose your hard drive and you don't have any backup you won't be able to recover.

To add a conclusion to this: there's always a cost somewhere!

--
David :wink:
Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs

Le 19 mai 2013 à 10:00, Ankit Jain ankitjaincs06@gmail.com a écrit :

Hi Alexander,

We don''t want to use replication, because replication will increase the storage cost.

Thanks,
Ankit Jain

On Friday, 17 May 2013 15:42:06 UTC+5:30, Alexander Reelsen wrote:

Hey,

I am sorry, but I dont understand you. Why do you refrain from using replication, even though it can solve your problem of high availability? Do you think replication is bad for you because of other reasons?
Maybe you can shed some light on this.

Thx!

--Alex

On Fri, May 17, 2013 at 11:16 AM, Ankit Jain ankitj...@gmail.com wrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

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 elasticsearc...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
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.

--
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.

6

Hi Ankit,

ES supports two features that are key to High Availbility : Replication and
Sharding.
You definitely want to use (even abuse :wink: both of them in an HA scenario.
Furthermore IMHO, disk storage is very cheap and there is no point or
benefit in not using replication.

my 2 cents

Stephane

On Friday, May 17, 2013 11:16:00 AM UTC+2, Ankit Jain wrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like
to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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.

7

I recommend to read
this: No such thing as a free lunch - Wikipedia
and further the following curl:

curl -s -XPUT 'localhost:9200/$your_index/_settings' -d '{
"index" : {
"number_of_replicas" : 1
}
}'

simon

On Sunday, May 19, 2013 2:21:24 PM UTC+2, Stephane Bastian wrote:

Hi Ankit,

ES supports two features that are key to High Availbility : Replication
and Sharding.
You definitely want to use (even abuse :wink: both of them in an HA scenario.
Furthermore IMHO, disk storage is very cheap and there is no point or
benefit in not using replication.

my 2 cents

Stephane

On Friday, May 17, 2013 11:16:00 AM UTC+2, Ankit Jain wrote:

Hi All,

We want to deploy Elasticsearch in production environment, we would like
to achieve high availability without using replication.

We have 10 ES storage nodes (0 replica), if one machine goes down, than
indices stored on that machine are not available for read.

How we can achieve read high availability, if any machines goes down?

Thanks,
Ankit

--
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.

8

I don't believe the request is for a free lunch. The request is to feel
full without having any food available at all.

:slight_smile:

On Monday, May 20, 2013 4:30:29 PM UTC-4, simonw wrote:

I recommend to read this:
No such thing as a free lunch - Wikipedia

--
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.