hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew Purtell <apurt...@apache.org>
Subject RE: Theoretical question...
Date Fri, 30 Apr 2010 14:52:17 GMT
Given your take, I encourage you to check out HBASE-1697.

   - Andy

On Fri Apr 30th, 2010 6:14 AM PDT Michael Segel wrote:

>
>Andrew,
>
>Not exactly.
>
>Within HBase, if you have access, you can do anything to any resource. I don't believe
there's a concept of permissions. (Unless you can use the HDFS permissions inside HBase...)
>
>So one idea was to isolate the hbase instance within the cloud.
>Since people talk about isolating hbase to different nodes than hadoop datanodes, this
kind of makes sense.
>
>I'm not a big fan of HOD with respect to virtualized clouds. And to be clear I mean taking
a physical box and splitting it in to virtualized servers. Running HOD on a large cloud of
100+ physical servers may have some value in a corporate cloud that is a shared resource.
>
>I'm sure there are folks at Sun, Dell, and IBM who would disagree with me, but when I
take a 8 core Intel box, and then split it in to two virtual 3 core boxes, I wonder if there
is going to be better performance than if I left it as a single 8 core box and ran more m/r
jobs?
>
>I realize that we have two issues for discussion. One is getting around the lack of security
and permissions within HBase, the other is virtualization.
>
>Both are interesting areas for discussion.
>
>Thx
>
>-Mike
>
>> Date: Thu, 29 Apr 2010 18:09:35 -0700
>> From: apurtell@apache.org
>> Subject: Re: Theoretical question...
>> To: hbase-user@hadoop.apache.org
>> 
>> > From: Michael Segel
>> 
>> > Imagine you have a cloud of 100 hadoop nodes.
>> > In theory you could create multiple instances of HBase on
>> > the cloud.
>> > Obviously I don't think you could have multiple region
>> > servers running on the same node.
>> > The use case I was thinking about if you have a centralized
>> > hadoop cloud and you wanted to have multiple developer
>> > groups sharing the cloud as a resource rather than building
>> > their own clouds.
>> 
>> This is somewhat like HOD (http://hadoop.apache.org/common/docs/current/hod_user_guide.html).
Have you looked at that?
>> 
>> > The reason for the multiple hbase instances is that you
>> > don't have a way of setting up multiple instances like
>> > different Informix or Oracle databases/schemas on the same
>> > infrastructure.
>> 
>> Right.
>> 
>> Well there is a simple (and under development, lightly tested as yet, etc.) multiuser
mode in Stargate that gives multiple users each the illusion of a private HBase instance while
sharing a common HBase cluster underneath. This is something I'll continue to work on as I
have time.
>> 
>> Also my employer is sponsoring development of HBASE-1697, and integration of HBase
into the secure version of Hadoop (http://bit.ly/75011o) that Yahoo is working on. For example
HBase would offer RBAC and might also use HDFS block tokens in a manner that allows you to
reason about user isolation down through the whole stack.
>> 
>> salesforce.com is a multitenant service built on a shared database infrastructure.
They've talked about their rationale for building their SaaS service this way. It's worth
it to Google a bit to find it and read. Partitioning cloud resources increases management
complexity and reduces the benefit of the cloud -- the efficiencies of scale. It's technically
possible to partition cloud resources but economically inefficient and suboptimal to do so.
>> 
>>    - Andy
>> 
>> 
>> 
>>       
>> 
> 		 	   		  
>_________________________________________________________________
>Hotmail is redefining busy with tools for the New Busy. Get more from your inbox.
>http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_2


      


Mime
View raw message