Re:Death of the Pure IaaS Cloud: Part 2-
by hapless on Monday April 04, 2011 @12:29PM
(#2407225)
I'm tired of hearing about the "death" of IaaS. This guy is paranoid. The reason that "PaaS" platforms are multiplying is that they are easy to put together. By deliberately restricting what the customer can do, you can offer cheaper and higher-performing services.
Vendor lock-in is a nice plus for the vendor, but that alone wouldn't drive these product offerings.
The truth is that we haven't worked out the kinks in infrastructure as a service. Only trivial designs can be implemented in today's public clouds. Judging from vCloud Service Director (or whatever Redwood is called this week), that's not likely to change in the immediate future.
Get back to me when "IaaS" products permit the kind of extensive network design customization that your managed service colo host will typically support. If those types of products are supplanted by vendors pushing PaaS, then I'll start to worry.
Re:Death of the Pure IaaS Cloud: Part 2-
by cloudsigma on Wednesday April 13, 2011 @03:43AM
(#2441187)
In terms of our platform:
* Customer-managed firewalls
Yes
* Redundant switching/interior routing
Yes
* Private layer 2 networks
Yes we have private VLAN functionality. We'll be expanding this to up to 8 per server shortly.
* Bring your own IP space
Yes, this needs to be manually added to our BGP sessions but isn't a problem and we do this regularly for customers.
* Redundant exterior routing (e.g. I advertise my prefix through a new provider during an extended downtime)
We have multiple redundant carriers and they failover automatically. Actually the failover happens instantaneously, even if you are actively pinging you don't even get one dropped packet :-) If you have your own private infrastructure space elsewhere we also are able to choose the preferred carrier too.
* Out-of-band access
Not sure in what context you are talking about here. If you extrapolate I can give feedback.
Vendor lock-in is a nice plus for the vendor, but that alone wouldn't drive these product offerings.
The truth is that we haven't worked out the kinks in infrastructure as a service. Only trivial designs can be implemented in today's public clouds. Judging from vCloud Service Director (or whatever Redwood is called this week), that's not likely to change in the immediate future.
Get back to me when "IaaS" products permit the kind of extensive network design customization that your managed service colo host will typically support. If those types of products are supplanted by vendors pushing PaaS, then I'll start to worry.
"extensive network design customization that your managed service colo host will typically support"
could you give a couple of examples of the sort of thing you are talking about?
* Redundant switching/interior routing
* Private layer 2 networks
* Bring your own IP space
* Redundant exterior routing (e.g. I advertise my prefix through a new provider during an extended downtime)
* Out-of-band access
* Customer-managed firewalls
Yes
* Redundant switching/interior routing
Yes
* Private layer 2 networks
Yes we have private VLAN functionality. We'll be expanding this to up to 8 per server shortly.
* Bring your own IP space
Yes, this needs to be manually added to our BGP sessions but isn't a problem and we do this regularly for customers.
* Redundant exterior routing (e.g. I advertise my prefix through a new provider during an extended downtime)
We have multiple redundant carriers and they failover automatically. Actually the failover happens instantaneously, even if you are actively pinging you don't even get one dropped packet :-) If you have your own private infrastructure space elsewhere we also are able to choose the preferred carrier too.
* Out-of-band access
Not sure in what context you are talking about here. If you extrapolate I can give feedback.