Over on the Enterprise Irregulars mailing list, someone asked what “private cloud” means. As ever, that group has a fascinating response. Here’s how I responded, which sums up a lot of what I’ve been saying on the topic of late, mostly in the IT Management & Cloud podcast:
What Private Cloud Seems to Mean
When I encounter “private cloud,” it usually means applying cloud technologies (virtualization, self-service, automation, generalized pools of resources [driven by the previous]) behind the firewall. It’s more than virtualization consolidation, though virtualization is a huge part – if not the most important technological enabler. The self-service (or “run book,” if you’re old school) parts are important as well.
I’m extremely suspicious of private cloud being much of a cost savings, but if it was done properly you would get the non-financial benefits of (public) cloud computing: quick boot-strapping, not having to wait 4 weeks for the DBA to get to your email, etc. In theory, this means IT can stop being slow (as they’re trained and rewarded to do with their “we fear change” philosophy – needed for how crappy enterprise IT tends to be during updates) and speed up. There’s also some (potential) process change in getting developers and operations people more friends, flying under the phrase of “dev/ops” in the gasbagosphere (myself included there).
The main thing that frustrates me about private cloud is that it seems to be a way for vendors to make money twice off cloud computing: once in the private cloud transformation, and then once all the fears blow over/get fixed, a second time for the public cloud transformation. It’s like retail in the early 90’s: remember when “no one” would dare type their credit card number into “the Internet”? Now everyone does, of course.
That cynicism aside, I do think that the happy path of “private cloud” is much better than whatever you want to call how IT is run now (ITSM, maybe, but that’s a bit unfair to ITSM). The main thing that frustrates me is this focus on “we must own our precious data” instead of innovating around/beyond that constraint. Sure, it’s “real,” and I’m not belittling it, but I expect more from the industry than to let that be a wall that keeps innovation at bay.
More
That’s an incomplete view, but I think it applies to the majority of instances of “private cloud” I come across. I’m sure the EI thread will pull out plenty more interesting to pile on. You can’t really be an absolutist about cloud computing, let along private clouds: throwing out a loose definition like the above is always a good honey-pot for better ones. For example, my podcasting buddy John Willis has a pretty good rant on why private cloud is actually awesome, ask him about it sometime. Also, see James’ discussion of EMC and VMWare’s private cloud recent machinations.
Extremely smart to be "suspicious of private cloud being much of a cost savings". In a mid sized business deployment, look at lost economies of scale as well as both hardware and software resources needed to get to an equivalent of a major cloud services offering; if anything, cost will increase.
Cloud computing cannot be private because it's about purchasing computing resources as one needs them, to scale up or down, from a massively scalable computing services environment, the kind of thing that renders current IT paradigms pretty much useless. Then, I’m still waiting for someone above my pay grade to bring clarity to cloud anything. My biggest frustration; private cloud thinking reinforces the status quo, inhibiting collaboration and change.
Les from my experience that is lots of spare compute and storage capacity sitting idle or severely under-utilized in todays IT environments you do not need to be a big as amazon in your offering to still make cost savings (time is money for some) by applying similar technologies, techniques and processes to a "private" cloud.
We are extremely suspicious about private clouds. Mainly because the main advocates of private clouds are those selling HW – Cisco, EMC, F5, IBM, etc. Virtualization has suddenly become = cloud. Self-service and elasticity are one aspect of cloud computing but they are far from the main benefits. The main benefits are being able to abstract away the management of HW/infrastructure platforms and getting access to platforms that keep getting better. For example, when I build on a platform such as Force.com, I automatically get to take advantage of the new UI, new reporting and social features, with no rewrites. That's the power of cloud computing, not eking out a little bit more savings than a virtualized datacenter! (more on our perspective here – http://blog.appirio.com/2009/08/amazon-vpc-privat…
Michael, I found it interesting that you noted suspicion that private clouds wouldn't get all the financial benefits (presuming you primarily meant economies of scale and no CapEx) of public clouds, yet you didn't mention all the other benefits that private clouds often lack — scalability on-demand, disaster recovery preparedness via ability to fail-over to multiple hosting locations (not RTO/RPO-style DR), the offloading of infrastructure management responsibility, just to name a few. As you say, virtualization is nice and a big part of private cloud, but doesn't address many of the operational enhancements that Cloud portends (or pretends depending on how sarcastic one feels at the moment).
You’re so awesome! I wont believe Ive read through anything such as this before. So awesome to discover someone with some original ideas on it all question. realy appreciate it to get starting this up. this web site is something which is required on the internet, anyone by having a little bit of creativity. useful work for carrying something which new to the web!