Hello,

 

we are currently evaluating HBase for a project. In respect to available
backup options, we found the following blog post here:

http://blog.sematext.com/2011/03/11/hbase-backup-options/

 

Probably well known for you guys here. So, am I right when I say that
there is no online (consistent backup while the system is in use) and
incremental backup utility we are used from the RDBMS world?

 

We would also like to try out various outage scenarios, e.g. pulling the
network cable out of one node or resetting the server etc ... while the
system is in use ... Anybody tried different outage scenarios with
HBase?

 

Thanks!

 

Thomas

 

Not sure what you read in Otis' bog but pretty ssure it's out of date.

Check out MapR stuff.

Sent from my Palm Pre on AT&T
On Jul 14, 2011 6:57 AM, Steinmaurer Thomas <[EMAIL PROTECTED]> wrote:

Hello,

 

we are currently evaluating HBase for a project. In respect to available

backup options, we found the following blog post here:

http://blog.sematext.com/2011/03/11/hbase-backup-options/

 

Probably well known for you guys here. So, am I right when I say that

there is no online (consistent backup while the system is in use) and

incremental backup utility we are used from the RDBMS world?

 

We would also like to try out various outage scenarios, e.g. pulling the

network cable out of one node or resetting the server etc ... while the

system is in use ... Anybody tried different outage scenarios with

HBase?

 

Thanks!

 

Thomas

 

To clarify what Mike means here, MapR supports HBase as well as atomic,
transactionally correct snapshots.  These snapshots allow point in time
recovery of the complete state of an HBase data set.

There is no performance hit when taking the snapshot and no maintenance
impact relative to HBase administration.  Snapshots can be scheduled or
invoked manually.

We also provide transactionally correct mirrors that have similar
characteristics but which allow actual copies to be made, possibly between
clusters.

This isn't quite related to HBase itself since MapR simply provides a
substrate on which to run HBase (and distributes the Apache released bits
for convenience).

On Thu, Jul 14, 2011 at 9:14 AM, Michael Segel <[EMAIL PROTECTED]>wrote:

> Not sure what you read in Otis' bog but pretty ssure it's out of date.
>
> Check out MapR stuff.
>
>
>
> Sent from my Palm Pre on AT&T
> On Jul 14, 2011 6:57 AM, Steinmaurer Thomas <[EMAIL PROTECTED]>
> wrote:
>
> Hello,
>
>
>
>
>
>
>
> we are currently evaluating HBase for a project. In respect to available
>
> backup options, we found the following blog post here:
>
>
>
> http://blog.sematext.com/2011/03/11/hbase-backup-options/
>
>
>
>
>
>
>
> Probably well known for you guys here. So, am I right when I say that
>
> there is no online (consistent backup while the system is in use) and
>
> incremental backup utility we are used from the RDBMS world?
>
>
>
>
>
>
>
> We would also like to try out various outage scenarios, e.g. pulling the
>
> network cable out of one node or resetting the server etc ... while the
>
> system is in use ... Anybody tried different outage scenarios with
>
> HBase?
>
>
>
>
>
>
>
> Thanks!
>
>
>
>
>
>
>
> Thomas
>
>
>
>
>
>
>
>
>
>
>

Otis' blog entry is from 4 months ago.  It still applies for the Apache
stack.

The Hbase book FAQ has a link to that blog for the backup question.

On 7/14/11 12:14 PM, "Michael Segel" <[EMAIL PROTECTED]> wrote:

>Not sure what you read in Otis' bog but pretty ssure it's out of date.
>
>Check out MapR stuff.
>
>
>
>Sent from my Palm Pre on AT&T
>On Jul 14, 2011 6:57 AM, Steinmaurer Thomas
><[EMAIL PROTECTED]> wrote:
>
>Hello,
>
>
>
>
>
>
>
>we are currently evaluating HBase for a project. In respect to available
>
>backup options, we found the following blog post here:
>
>
>
>http://blog.sematext.com/2011/03/11/hbase-backup-options/
>
>
>
>
>
>
>
>Probably well known for you guys here. So, am I right when I say that
>
>there is no online (consistent backup while the system is in use) and
>
>incremental backup utility we are used from the RDBMS world?
>
>
>
>
>
>
>
>We would also like to try out various outage scenarios, e.g. pulling the
>
>network cable out of one node or resetting the server etc ... while the
>
>system is in use ... Anybody tried different outage scenarios with
>
>HBase?
>
>
>
>
>
>
>
>Thanks!
>
>
>
>
>
>
>
>Thomas
>
>
>
>
>
>
>
>
>
>

On Thu, Jul 14, 2011 at 4:57 AM, Steinmaurer Thomas
<[EMAIL PROTECTED]> wrote:
> We would also like to try out various outage scenarios, e.g. pulling the
> network cable out of one node or resetting the server etc ... while the
> system is in use ... Anybody tried different outage scenarios with
> HBase?
>

Yes, this has been tried.  If you find anything, please let us know.
St.Ack

I should clarify.

4 months ago is 'old' with respect to the rate of change within Hadoop/HBase. (Its a scary good thing.)

Ted is correct that I'm looking at MapRTech's ability to do fast snap shots.

Its important to note that while the Apache stack has certain limitations, derivative works may not have the same limitations.  So while its important to look at the FAQ, its only a starting place.
While I have no skin in the Hadoop stack game, I do have a vested interest in recommending the best derivative work that meets my customers' requirements.

I'm not sure when and what HortonWorks is going to offer. They are really too new to have everything worked out. (And I do look forward to their support offering.)
When I spoke to their COO he indicated that they will be supporting the Apache release and selling a commercial support license for the Apache release, that doesn't mean that they will not end up with a derivative work in the near future.
(I'd actually love to have him or someone from HortonWorks come to Chicago in Sept to talk to our CHUG. ;-)
I'm sorry that I digressed but the point I was trying to make was that customers want a commercial support package and only derivative works meet this requirement. The ability to better manage the cluster, its resources, and provide better control over job management are critical features.

I really don't want to suggest that the FAQ start tracking all of the features in the derivative works, but if its going to reference outside blogs... maybe it should? (Or also link to their blogs?)

On a side note... IMHO I love the fact that we have a couple of companies doing commercial releases. The competition is good in that improvements are starting to happen at a faster pace.

-Mike

> From: [EMAIL PROTECTED]
> To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
> Date: Thu, 14 Jul 2011 13:39:41 -0400
> Subject: Re: HBase backup and outage scenarios in practice?
>
>
> Otis' blog entry is from 4 months ago.  It still applies for the Apache
> stack.
>
> The Hbase book FAQ has a link to that blog for the backup question.
>
>
>
> On 7/14/11 12:14 PM, "Michael Segel" <[EMAIL PROTECTED]> wrote:
>
> >Not sure what you read in Otis' bog but pretty ssure it's out of date.
> >
> >Check out MapR stuff.
> >
> >
> >
> >Sent from my Palm Pre on AT&T
> >On Jul 14, 2011 6:57 AM, Steinmaurer Thomas
> ><[EMAIL PROTECTED]> wrote:
> >
> >Hello,
> >
> >
> >
> >
> >
> >
> >
> >we are currently evaluating HBase for a project. In respect to available
> >
> >backup options, we found the following blog post here:
> >
> >
> >
> >http://blog.sematext.com/2011/03/11/hbase-backup-options/
> >
> >
> >
> >
> >
> >
> >
> >Probably well known for you guys here. So, am I right when I say that
> >
> >there is no online (consistent backup while the system is in use) and
> >
> >incremental backup utility we are used from the RDBMS world?
> >
> >
> >
> >
> >
> >
> >
> >We would also like to try out various outage scenarios, e.g. pulling the
> >
> >network cable out of one node or resetting the server etc ... while the
> >
> >system is in use ... Anybody tried different outage scenarios with
> >
> >HBase?
> >
> >
> >
> >
> >
> >
> >
> >Thanks!
> >
> >
> >
> >
> >
> >
> >
> >Thomas
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
>
     

On Fri, Jul 15, 2011 at 7:39 AM, Michael Segel
<[EMAIL PROTECTED]> wrote:
> I really don't want to suggest that the FAQ start tracking all of the features in the derivative works, but if its going to reference outside blogs... maybe it should? (Or also link to their blogs?)
>

That sounds like a big job tracking the differences and our doc
doesn't seem to be the right place for the table of feature xs and
ticks.

We could add a link to this page up on hadoop wiki?
http://wiki.apache.org/hadoop/Distributions%20and%20Commercial%20Support
(This page seems to be missing MapR).

Perhaps something similar on our wiki for hbase specific services and
support that is maintained by the interested parties rather?

St.Ack

> On a side note... IMHO I love the fact that we have a couple of companies doing commercial releases. The competition is good in that improvements are starting to happen at a faster pace.
>
> -Mike
>
>> From: [EMAIL PROTECTED]
>> To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
>> Date: Thu, 14 Jul 2011 13:39:41 -0400
>> Subject: Re: HBase backup and outage scenarios in practice?
>>
>>
>> Otis' blog entry is from 4 months ago.  It still applies for the Apache
>> stack.
>>
>> The Hbase book FAQ has a link to that blog for the backup question.
>>
>>
>>
>> On 7/14/11 12:14 PM, "Michael Segel" <[EMAIL PROTECTED]> wrote:
>>
>> >Not sure what you read in Otis' bog but pretty ssure it's out of date.
>> >
>> >Check out MapR stuff.
>> >
>> >
>> >
>> >Sent from my Palm Pre on AT&T
>> >On Jul 14, 2011 6:57 AM, Steinmaurer Thomas
>> ><[EMAIL PROTECTED]> wrote:
>> >
>> >Hello,
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >we are currently evaluating HBase for a project. In respect to available
>> >
>> >backup options, we found the following blog post here:
>> >
>> >
>> >
>> >http://blog.sematext.com/2011/03/11/hbase-backup-options/
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >Probably well known for you guys here. So, am I right when I say that
>> >
>> >there is no online (consistent backup while the system is in use) and
>> >
>> >incremental backup utility we are used from the RDBMS world?
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >We would also like to try out various outage scenarios, e.g. pulling the
>> >
>> >network cable out of one node or resetting the server etc ... while the
>> >
>> >system is in use ... Anybody tried different outage scenarios with
>> >
>> >HBase?
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >Thanks!
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >Thomas
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>>
>

On 15/07/11 16:39, Michael Segel wrote:
>
> I should clarify.
>
> 4 months ago is 'old' with respect to the rate of change within Hadoop/HBase. (Its a scary good thing.)
>
> Ted is correct that I'm looking at MapRTech's ability to do fast snap shots.
>
> Its important to note that while the Apache stack has certain limitations, derivative works may not have the same limitations.  So while its important to look at the FAQ, its only a starting place.
> While I have no skin in the Hadoop stack game, I do have a vested interest in recommending the best derivative work that meets my customers' requirements.
>

Not really a "backup/snapshot" scenario, but hbase replication is
documented : http://hbase.apache.org/replication.html

Maybe related to https://issues.apache.org/jira/browse/HBASE-1295 Multi
data center replication ?

> I'm not sure when and what HortonWorks is going to offer. They are really too new to have everything worked out. (And I do look forward to their support offering.)
> When I spoke to their COO he indicated that they will be supporting the Apache release and selling a commercial support license for the Apache release, that doesn't mean that they will not end up with a derivative work in the near future.
> (I'd actually love to have him or someone from HortonWorks come to Chicago in Sept to talk to our CHUG. ;-)
> I'm sorry that I digressed but the point I was trying to make was that customers want a commercial support package and only derivative works meet this requirement. The ability to better manage the cluster, its resources, and provide better control over job management are critical features.
>
> I really don't want to suggest that the FAQ start tracking all of the features in the derivative works, but if its going to reference outside blogs... maybe it should? (Or also link to their blogs?)
>
> On a side note... IMHO I love the fact that we have a couple of companies doing commercial releases. The competition is good in that improvements are starting to happen at a faster pace.
>
> -Mike
>
>> From: [EMAIL PROTECTED]
>> To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
>> Date: Thu, 14 Jul 2011 13:39:41 -0400
>> Subject: Re: HBase backup and outage scenarios in practice?
>>
>>
>> Otis' blog entry is from 4 months ago.  It still applies for the Apache
>> stack.
>>
>> The Hbase book FAQ has a link to that blog for the backup question.
>>
>>
>>
>> On 7/14/11 12:14 PM, "Michael Segel"<[EMAIL PROTECTED]>  wrote:
>>
>>> Not sure what you read in Otis' bog but pretty ssure it's out of date.
>>>
>>> Check out MapR stuff.
>>>
>>>
>>>
>>> Sent from my Palm Pre on AT&T
>>> On Jul 14, 2011 6:57 AM, Steinmaurer Thomas
>>> <[EMAIL PROTECTED]> wrote:
>>>
>>> Hello,
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> we are currently evaluating HBase for a project. In respect to available
>>>
>>> backup options, we found the following blog post here:
>>>
>>>
>>>
>>> http://blog.sematext.com/2011/03/11/hbase-backup-options/
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> Probably well known for you guys here. So, am I right when I say that
>>>
>>> there is no online (consistent backup while the system is in use) and
>>>
>>> incremental backup utility we are used from the RDBMS world?
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> We would also like to try out various outage scenarios, e.g. pulling the
>>>
>>> network cable out of one node or resetting the server etc ... while the
>>>
>>> system is in use ... Anybody tried different outage scenarios with
>>>
>>> HBase?
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> Thanks!
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> Thomas
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>
>    
--
Eric Charles
http://about.echarles.net

Hola,

If there is indeed something out out of date there, please do let me know - I'd like to keep that post up to date, esp. since it's linked from the HBase Book. 

Thanks,
Otis
----
Sematext :: http://sematext.com/ :: Solr - Lucene - Hadoop - HBase
Hadoop ecosystem search :: http://search-hadoop.com/

>________________________________
>From: Michael Segel <[EMAIL PROTECTED]>
>To: "[EMAIL PROTECTED]" <[EMAIL PROTECTED]>
>Sent: Thursday, July 14, 2011 12:14 PM
>Subject: Re: HBase backup and outage scenarios in practice?
>
>Not sure what you read in Otis' bog but pretty ssure it's out of date.
>
>Check out MapR stuff.
>
>
>
>Sent from my Palm Pre on AT&T
>On Jul 14, 2011 6:57 AM, Steinmaurer Thomas <[EMAIL PROTECTED]> wrote:
>
>Hello,
>
>
>
>
>
>
>
>we are currently evaluating HBase for a project. In respect to available
>
>backup options, we found the following blog post here:
>
>
>
>http://blog.sematext.com/2011/03/11/hbase-backup-options/
>
>
>
>
>
>
>
>Probably well known for you guys here. So, am I right when I say that
>
>there is no online (consistent backup while the system is in use) and
>
>incremental backup utility we are used from the RDBMS world?
>
>
>
>
>
>
>
>We would also like to try out various outage scenarios, e.g. pulling the
>
>network cable out of one node or resetting the server etc ... while the
>
>system is in use ... Anybody tried different outage scenarios with
>
>HBase?
>
>
>
>
>
>
>
>Thanks!
>
>
>
>
>
>
>
>Thomas
>
>
>
>
>
>
>
>
>
>
>
>
>