openstack team mailing list archive
Mailing list archive
Re: Announcing Elastic Hadoop on OpenStack project
First of all, thank you for the question.
EHO is not just a Hadoop deployment or management tool, but complete fully-integrated native OpenStack component. EHO will provide an ability to apply different deployment and provisioning tools using extension mechanism. So, deployment could be backed by Chef, Puppet, JuJu, etc or even you can use handmade provisioning tools. The same thing is with management and monitoring tools.
As for integration with OpenStack, for example, EHO web GUI is a part of OpenStack Dashboard (preview: http://eho.mirantis.com/horizon/index.html, screencast will be available in a few days). Additionally, EHO uses OpenStack Identity Service (Keystone) for auth routines. All resources (VMs, IPs, volumes) used for cluster creation are provisioned by EHO.
Generally, we build complete extensible solution with UI and REST API available right away - not just a toolbox.
On Mar 13, 2013, at 23:35, Brad Knowles <bknowles@xxxxxxxxxxxxxx> wrote:
> On Mar 13, 2013, at 1:16 PM, Sergey Lukjanov <slukjanov@xxxxxxxxxxxx> wrote:
>> The main goal of this initiative, codenamed EHO (Elastic Hadoop on OpenStack), is to create native OpenStack component for fast provisioning and operating Hadoop clusters on demand.
> I'm curious -- how does this compare to IronFan from the folks at InfoChimps? Seems to me like that's a complete "Hadoop in a Box" type of solution, based on Opscode Chef plus a bunch of other components.
> Brad Knowles <bknowles@xxxxxxxxxxxxxx>
> Senior Consultant