The private cloud debate is building up steam, but is it worth having?

Slowly but steadily the debate in the blogosphere about private clouds is increasing. Now it is always good to see some debate, but is this a debate worth having? Will the cloud long term not be about other things than who owns a machine?

Under provocative titles like “Private cloud discredited, part 1”  and “Do We Really Need Private Clouds?”  the private cloud debate is building up steam. The first blog is actually called “part 1” because the author is sure there will be a part two, given the raging emotions and all the opinions being aired.
The second one is part of a very readable guest series by IT analyst avant la lettre Robin Bloor at Cloud Commons. Cloud Commons  is a cloud consumer rating service community site, like and but for cloud products and services, that CA Technologies helped initiate.
Now it is always good to see debate, and I vividly remember when we all got exited some years ago about Open Systems (with Open roughly being defined as anything running on Unix versus anything that was not running on Unix- including mainframes, AS/400’s, HP3000’s, etc. etc.). To be honest , that debate was maybe as productive as a debate about private versus public clouds may turn out to be. In my view the most important thing that the cloud can bring is namely that cloud (finally) decouples the application from the underlying infrastructure. As a result it matters a lot less where it runs (private or public). In a comment on Robin’s “Do We Really Need Private Clouds?” blog Jonathan Davis, CTO of DNS Europe, introduces a good example of that principle. Using a cloud platform (Applogic* in this case) his company makes it possible that applications can be deployed transparently and instantly over grids of compute capacity (re. discussion of grids, see Robin’s earlier post in the mentioned series) regardless of whether these clouds are private (hosted or internal), public or a combination (hybrid).

Where to start
The remaining Private versus Public question then is: where to start. Do you start with less sensitive applications on a public cloud and then expand what you learned to core apps onto maybe a private cloud, or v.v. do you start with a more sensitive app on a private cloud and expand to public when you feel that is proven and secure enough for that application.

Surprisingly (at least to me) there was some very clear guidance given in the cloud scenario session at last month’s annual Gartner Symposium/ITxpo**. Basically a kind of “comply or explain” approach was suggested there. It basically suggested: first explore whether a job can be done with a public cloud (“comply”), and only if there are valid and severe reasons to not go public (“explain”) than consider private. I’m paraphrasing so check with your analyst for exact wording and/or check out the free videos recordings of this year’s symposium at  . Unfortunately the break-out sessions on cloud are not available for free there, but it does offer all plenary keynotes, all industry CEO interviews (Bennioff, Chambers, Ballmer, Dell) and … a recording of my session ;-).

During the symposium, Gartner also indicated that security concerns should be more seen as valid but temporary challenges to be addressed and overcome, rather than as a reason (or excuse) to discard public clouds. At last weeks Datacenter Summit one of Gartner’s lead analysts on cloud, Thomas Bittman, gave a slightly more nuanced view. Understandable as this wa sthe week of Thankgiving and the folks in the room were predominantly the guys running today’s private datacenters (Thanksgiving –> Turkeys 🙂 ). He highlighted some scenario’s where private clouds makes perfect sense (eg. stable, predictable loads). In his worthwhile and balanced session he also noted that the current emphasis on Infrastructure as a Service (where the Private versus Public debate mainly plays today) over Platform and Software as a Services comes from the fact that IaaS can run today’s existing applications and does not have to wait for a next generation of apps, as developing such new applications simply takes time.

A new generation of cloud applications
In my view this new generation applications will be very different from the applications we run today, which makes it even more important that these new generations of applications will no longer be tight to underlying infrastructures. New generations come with new applications, mainframes introduced OLTP, Mini’s or distributed systems (both open and proprietary) introduced departmental systems and later packaged applications like MRP and ERP and internet web systems introduced the age of e-commerce, where we started buying books and gadgets online and doing our banking on-line (remember when everything was called e-something). So “re-hosting” our existing apps to (either private or public) clouds is only a very small part of the long term cloud story. Not that next year, as I will address in my upcoming #IT2011 predictions, won’t be a very lucrative year for many vendors – including the one I work for – helping organizations move their existing applications to (hosted or internal) private clouds. Not to mention that 2011 is likely to be the year of the IaaS KillerApp TestDev-clouds, that make test/dev machines (a whopping 70% of the average IT organization’s machine park) available in a much more flexible, economical and ecological way. My college Marv Waschke wrote about this last week as a perfect way to gain experience with private and public cloud scenario’s 

But the big long term story in my view is that cloud will be ideal for a generation of new applications. Applications that allow organizations to collaborate with other organizations – so not the now much talked about in-company twitter and facebook clones that enable people to waste as much time at the office as they do at home. Through these new collaboration applications, organizations can take business processes that were traditionally done in house and source them as “as a service”. These processes can vary from bill collecting, invoicing, physical distribution, repair handling and HR to full manufacturing or product design. Having companies be able to specialize and offer these services to many organizations will enable them to achieve massive economies of scale. Note that many of these services which will be largely or completely information/software based. An example: Imagine the efficiencies of one company handling repairs for several large mobile phone manufacturers versus each company having to arrange their repairs themselves. Most phone manufacturers sell through the same resellers, use the same repair centers, source form the same Chinese factories, etc. etc. Hooking these up ones to a central platform used by multiple players can give an enormous platform effect. An early European player in the area is (who BTW also run a cloud academy, but specific about these business aspects which they call “Cloud Processes for the Value Network”. At their site their head of product marketing also shares a handy list of 10 essential elements to create and run such a next generation cloud service). These types of new generation cloud applications will render efficiencies far beyond any pure IT savings or efficiencies imaginable (see also my earlier entry re. Gitex)

Now you may say that companies have started this move towards specialization and outsourcing of processes already some years ago. And you’re right, but so far they did so often despite of support in IT applications. Thanks to the cloud, IT can now become the big promoter, enabler and catalyst for this. In fact, I came across the scenario of “repair handling as a service” over a decade ago, while introducing XML for a previous employer. But back then the idea of bringing crucial functions outside the firewall and outside the realm of internal IT was just a bit too revolutionary. The fast growing acceptance of cloud computing as a model (often even more on the business side than on the IT side) is rapidly changing this. And to be honest, public cloud may clearly have an advantage over private cloud for such public services, as it is already located outside single companies proverbial firewalls.

PS many links in this one, if you read only one, make it Robin Bloor’s series at Cloud Commons, his havemacwillblog was one of the first IT blogs and this new series shows that experience does count.

*Disclosure: The cloud platform Applogic is a platform by 3Tera, now part of CA Technologies, my employer. DNS Europe spoke at the invitation of CA about their experiences with Applogic at VM World in Kopenhagen.
**CA Technologies is a premier sponsor of Gartner’s Symposium/ITxpo and Datacenter Summit.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: