Which RAID config for ES?

So given the built in fault tolerance of Elasticsearch across the cluster
are people adventurous enough to use RAID0?

I'm thinking of middle ground like RAID5...

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/f57383fd-a25c-4be9-a04a-ed391fddcbcf%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

On Thu, Sep 25, 2014 at 9:58 AM, John Smith java.dev.mtl@gmail.com wrote:

So given the built in fault tolerance of Elasticsearch across the cluster

are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on any
more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD


Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett nik9000@gmail.com wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith java.dev.mtl@gmail.com
wrote:

So given the built in fault tolerance of Elasticsearch across the cluster

are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on any
more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAKdsXoG4iuN%3DQtmrnSiFFTfZjk3RmoyJztynw4qebi0bUSxe7A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

So all your machines just have 2 drives each? Isn't that expensive?

On Thursday, 25 September 2014 10:01:42 UTC-4, Nikolas Everett wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith <java.d...@gmail.com
<javascript:>> wrote:

So given the built in fault tolerance of Elasticsearch across the cluster

are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on any
more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/b3650a09-dd36-455b-b882-6b8d4ca47257%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Not so much concerened about speed I know that RAID0 will give some
performance boost.

Just curious how tolerant an ES cluster is on RAID 0.

On Thursday, 25 September 2014 10:12:10 UTC-4, Jörg Prante wrote:

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD

http://core0.staticworld.net/images/article/2014/06/intel_ssd_raid-100315228-orig.png

Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett <nik...@gmail.com
<javascript:>> wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith <java.d...@gmail.com
<javascript:>> wrote:

So given the built in fault tolerance of Elasticsearch across the cluster

are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on
any more then two disks just feels squicky.

Nik

--
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:>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/46d0090a-45e0-464a-9d31-2fc1175d30b0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Not so much concerned about the performance boost of RAID 0 but rather how
fault tolerant is ES on RAID 0.

On Thursday, 25 September 2014 10:12:10 UTC-4, Jörg Prante wrote:

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD

http://core0.staticworld.net/images/article/2014/06/intel_ssd_raid-100315228-orig.png

Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett <nik...@gmail.com
<javascript:>> wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith <java.d...@gmail.com
<javascript:>> wrote:

So given the built in fault tolerance of Elasticsearch across the cluster

are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on
any more then two disks just feels squicky.

Nik

--
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:>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/9a2f5e2b-865d-443f-91aa-20cc267c07d5%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

So assuming 2 node cluster. If you lose the array on one machine, you still
have the other.

I guess it comes down to what are the chances that multiple RAID0 arrays
fail at the same time on the cluster

On Thursday, 25 September 2014 13:47:13 UTC-4, John Smith wrote:

Not so much concerned about the performance boost of RAID 0 but rather how
fault tolerant is ES on RAID 0.

On Thursday, 25 September 2014 10:12:10 UTC-4, Jörg Prante wrote:

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD

http://core0.staticworld.net/images/article/2014/06/intel_ssd_raid-100315228-orig.png

Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett nik...@gmail.com
wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith java.d...@gmail.com wrote:

So given the built in fault tolerance of Elasticsearch across the

cluster are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on
any more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/986ed9fb-82da-41d9-b0fb-6a999dde1c29%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

If you have replica > 1, more than one server is allowed to fail.

SSDs, also enterprise versions (you can recognize them by the price tag),
have erasing limits - enterprise SSD live longer. When they exceed the
limit, they tend to die abruptly without warning. This is a bit different
from spindle disks with all the SMART monitoring in place where controllers
can intervene. So if an SSD dies, the server is decommissioned for
maintenance, for replacing the faulty SSD(s).

Here are some graphs: https://ef.gy/statistics:ssd-write-endurance

Jörg

On Thu, Sep 25, 2014 at 8:08 PM, John Smith java.dev.mtl@gmail.com wrote:

So assuming 2 node cluster. If you lose the array on one machine, you
still have the other.

I guess it comes down to what are the chances that multiple RAID0 arrays
fail at the same time on the cluster

On Thursday, 25 September 2014 13:47:13 UTC-4, John Smith wrote:

Not so much concerned about the performance boost of RAID 0 but rather
how fault tolerant is ES on RAID 0.

On Thursday, 25 September 2014 10:12:10 UTC-4, Jörg Prante wrote:

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD

http://core0.staticworld.net/images/article/2014/06/intel_
ssd_raid-100315228-orig.png

Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett nik...@gmail.com
wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith java.d...@gmail.com
wrote:

So given the built in fault tolerance of Elasticsearch across the

cluster are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on
any more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit https://groups.google.com/d/
msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%
2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/986ed9fb-82da-41d9-b0fb-6a999dde1c29%40googlegroups.com
https://groups.google.com/d/msgid/elasticsearch/986ed9fb-82da-41d9-b0fb-6a999dde1c29%40googlegroups.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAKdsXoHrSZs2xsO2T%2BZzDxywbA_Ta0dVkNZAVpW_qQA9RfA_sA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Raid 0 , don't use 5 as the performance is sucky and ES is i/o intensive.

Regards,
Mark Walkom

Infrastructure Engineer
Campaign Monitor
email: markw@campaignmonitor.com
web: www.campaignmonitor.com

On 26 September 2014 06:01, joergprante@gmail.com joergprante@gmail.com
wrote:

If you have replica > 1, more than one server is allowed to fail.

SSDs, also enterprise versions (you can recognize them by the price tag),
have erasing limits - enterprise SSD live longer. When they exceed the
limit, they tend to die abruptly without warning. This is a bit different
from spindle disks with all the SMART monitoring in place where controllers
can intervene. So if an SSD dies, the server is decommissioned for
maintenance, for replacing the faulty SSD(s).

Here are some graphs: https://ef.gy/statistics:ssd-write-endurance

Jörg

On Thu, Sep 25, 2014 at 8:08 PM, John Smith java.dev.mtl@gmail.com
wrote:

So assuming 2 node cluster. If you lose the array on one machine, you
still have the other.

I guess it comes down to what are the chances that multiple RAID0 arrays
fail at the same time on the cluster

On Thursday, 25 September 2014 13:47:13 UTC-4, John Smith wrote:

Not so much concerned about the performance boost of RAID 0 but rather
how fault tolerant is ES on RAID 0.

On Thursday, 25 September 2014 10:12:10 UTC-4, Jörg Prante wrote:

If you want speed, use RAID 0.

I have 4 cheap SSDs on one SAS-2 controller in RAID 0.

Example for SSD

http://core0.staticworld.net/images/article/2014/06/intel_
ssd_raid-100315228-orig.png

Jörg

On Thu, Sep 25, 2014 at 4:01 PM, Nikolas Everett nik...@gmail.com
wrote:

On Thu, Sep 25, 2014 at 9:58 AM, John Smith java.d...@gmail.com
wrote:

So given the built in fault tolerance of Elasticsearch across the

cluster are people adventurous enough to use RAID0?

Absolutely. We only do it with pairs of disks though because RAID0 on
any more then two disks just feels squicky.

Nik

--
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.
To view this discussion on the web visit https://groups.google.com/d/
msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%
2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0pb5PiOE9Je8nG3mB5qq%2BEvk_Tnh4k06FXUd%2BestPmmg%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/986ed9fb-82da-41d9-b0fb-6a999dde1c29%40googlegroups.com
https://groups.google.com/d/msgid/elasticsearch/986ed9fb-82da-41d9-b0fb-6a999dde1c29%40googlegroups.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CAKdsXoHrSZs2xsO2T%2BZzDxywbA_Ta0dVkNZAVpW_qQA9RfA_sA%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CAKdsXoHrSZs2xsO2T%2BZzDxywbA_Ta0dVkNZAVpW_qQA9RfA_sA%40mail.gmail.com?utm_medium=email&utm_source=footer
.

For more options, visit https://groups.google.com/d/optout.

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAEM624asZGLyGuOAjZ9%2B_PENqxy4MxD1EiQOnTqRMBwTNTJ0aQ%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

1 Like