Open Stack

On Mon, Jul 11, 2011 at 12:27:16PM +0200, Soren Hansen wrote:
> 2011/7/11 Ewan Mellor <Ewan.Mellor at eu.citrix.com>:
> > No, not string vs guid.  Current AWS IDs are 32 bits.  Being a small key
> > space, this means that you either need to allocate them incrementally
> > (implying a distributed transaction across the incrementer)
> 
> This is only a real problem if you insist on generating them in real
> time rather than pre-allocate them. Each compute node could have pool
> of thousands of ID's it could use as it pleased. That would still
> allow for millions of compute nodes. The ID's could be centrally
> assigned. Even if the central component that hands them out goes away,
> a couple of thousand ID's should provide ample time for a replacement
> to be spun up (or for the original one to come back).

This assumes you still have something you can call "central". What
about hybrid or peer-to-peer clouds? Does each zone get assigned a
ID allocator service? Each account? What if it is behind a firewall
but I have public cloud bursting and don't want to open it up?

I think we need to use a completely distributed solution and just
let EC2 be second class, requiring a local, centralized mapping
service per endpoint (or endpoint cluster) if you really want to use
it. This will provide more incentive for folks to move to the OS API,
support the OS API in popular tools and services, and develop new
(and hopefully better) tools for it.

Just my thoughts,
-Eric