openstack team mailing list archive

[William DeCoste <wdecoste@xxxxxxxxxx>]

Hi all,

Apologies for coming in so late to this conversation. Thought I'd 
mention the current capability that Red Hat's OpenShift PaaS provides 
for application auto-scaling. A user can choose to create a scaled 
application which will:

1) Create a primary gear 
(https://openshift.redhat.com/community/faq#t6n11272) which will run an 
instance of the application stack and an HAProxy load-balancer.
2) Create a secondary gear running another instance of the application stack
3) HAProxy will load-balance between the primary and secondary gear(s)
4) As load increases/descreases through HAProxy, OpenShift will 
automatically spin up/down additional secondary gears.

OpenShift does not require the HEAT API, but this could be used to spin 
up the various machines needed to make a HA and stable OpenShift 
environment. Also, when nodes run out of resources to host new 
applications, we could use the HEAT API to spin up new nodes.

Thanks -Bill

 * **To**: openstack@xxxxxxxxxxxxxxxxxxx <mailto:openstack@DOMAIN.HIDDEN>
 * **From**: Frans Thamura <frans@xxxxxxxxxxxx
   <mailto:frans@DOMAIN.HIDDEN>>
 * **Date**: Fri, 26 Oct 2012 12:46:49 +0700

Hi All

Anyone can give me reference, related to scaling PaaS system in OpenStack?

how (more basic better) scalable is implementing PaaS in OpenStack?

right now, we create virtual machine and install ubuntu inside, and
run CloudFoundry or OpenShift to make it PaaS enable.

my target for PaaS is to run our Java apps inside cloud environment.

in another world, we have Liquid VM, but it is not opensource yet,
part of Java VE Virtual Edition. The JVM can boot direct from the
hypervisor.

I still researching the theory behind scalability of cloud esp in
openstack + cloudfoundry.

F

--
Bill DeCoste
Principal Software Engineer, Red Hat
978-204-0920
wdecoste@xxxxxxxxxx