Tuesday, June 30, 2009

Our cloud offering is ... a stamp.

I was all excited about Red Hat's entry into the cloud computing space until I read the press release.
Here's my problem ...

Today, a whole range of public cloud providers offer Red Hat as a virtual machine operating system (along with every other operating system). Tomorrow, Red Hat will certify some of those cloud providers.

Either I'm missing the point, or this is just the same as before but with a stamp of approval from Red Hat? Is it me or does that just seem incredibly arrogant? We can fix the cloud by applying our magic pixie dust stamp of approval.

I've worked extremely hard with the Ubuntu team and Eucalyptus to provide users with their own open source cloud system for building private clouds. We've created official images which run both on your private cloud and a public provider (i.e. Amazon EC2) and an ecosystem of management tools (RightScale, CohesiveFT).

We've provided real products for real users and our aim has been to provide an open source reference model for cloud computing at the infrastructure layer of the stack.

Whilst I agree that the solution to many of enterprise concerns over cloud can be resolved by such a common substrate, this needs to be far more than just saying run our operating system in your VMs and here's our stamp of approval.

I'm seriously disappointed. However, disappointment collapses into despair at sight of the following statement:

"How can a CIO be sure that an application written for a Google cloud will work with Salesforce, Amazon, or another cloud?"

[Hint: You can't, they're completely different levels of the computing stack]

I keep on getting this horrible feeling that they haven't got to grips with the changes in the environment but that doesn't make sense because the changes are plainly visible. Every time I look at the landscape it gets me confused ... what on earth are they upto? Have they found some new area of value I just don't get?

Monday, June 22, 2009

Black is the new ... black.

My friend James is reported to have said :-

"as far as the enterprise is concerned, cloud [computing] is the new VMWare, and IBM wants a fat slice this time".

According to another friend of mine, James Urquhart, it looks like the new slice IBM is getting is based upon .... VMWare.

Numb Nuts ...

Numb Nuts are those people whose mental faculties have somewhat been pacified by the onset of an easy culture. It would be unfair to describe them as feeble-minded as it's more about an unwillingness to think.

In the "cloud" world there are many examples of second sourcing failure through proprietary technology (e.g. Zimki, CogHead & Virtual Iron). Equally there are many Numb Nuts who still believe that proprietary is the way forward.

A femtogram of thought would show that in the service world, competition and value should be around services whilst the bits are free.

If you're building a cloud or using a cloud, ask yourself where is the open source reference implementation of this? Where are my second sourcing options?

If you don't have this, then check your face in the mirror each day and look for the tell-tale sign of "Numb Nut" appearing in bold lettering across your forehead.

As the habitual users of Virtual Iron have discovered, it'll happen soon enough.

[El Reg talking about Virtual Iron]
"anyone that built their hosting infrastructure on VI...is now totally in the shit".

[MasterMark speaking about Coghead 'going tits up']
"that left its customers essentially in the unfortunate position of being 'shit out of luck'.

P.S. Before anyone whines that Amazon EC2 is proprietary - try Eucalyptus, Globus Nimbus or any of the other open source re-implementations of the API.

P.P.S. Before anyone huffs that their proprietary system is backed by a big company, in my view "financial safety" is either open sourced or much bigger than Lehman Brothers i.e. revenues above $20 billion and a headcount higher than 28,000 - 'nuff said.

Monday, June 15, 2009

The cloud isn't just vapour.

I was recently asked whether I thought cloud was vapour, whether open sourced cloud systems could solve many of the adoption concerns over cloud computing, whether it was likely that such systems would appear, what standards would they support and which distribution would make the first move? I know I haven't been blogging much recently but I was surprised by the questions.

First, regardless of whatever definition you use to describe cloud computing (a fairly hopeless task in my view), the term merely identifies an underlying shift of I.T. from a product to a service based economy. It's a consequence of :-
  1. Certain I.T. activities becoming suitable for service provision through volume operations (i.e. those activities are well defined and ubiquitous
  2. The existence of mature enough technology to support this (Popek et al wrote the book on virtualisation back in 1974)
  3. A change in business attitude towards I.T. (Carr and others, namely Strassmann, pointed out that much of I.T. is simply seen as a cost of doing business.)
  4. The concept of utility computing provision (i.e. the provision of suitable I.T. resources much like other utility providers, as forecast by McCarthy back in the 1960's.)
Take away any of these elements and cloud computing wouldn't represent the upheaval and the disruption that it does today. As for trying to precisely define it, try first coming up with a short and punchy description of industrial revolution without hand-waving and referring to numerous tomes. The problem with cloud computing is that it isn't one thing, it's a transition caused by many factors.

I do believe that open source reference models (or what I used to call open sourced standards) are key to the development of the cloud industry because of the second sourcing concerns of enterprises. I strongly believe that private and hybrid clouds (using both private and public resources) will help develop this industry in the short term. Open source should also dominate this change as it is the only viable route to utility computing marketplaces with competition based upon services rather than lock-in. I've not changed my tune since 2006, I see no reason to change now.

During this time of transition (which is after all what is happening) standards will be incredibly important. Despite all the noise we already have a defacto standard at the infrastructure layer of the computing stack - it's called the Amazon EC2 API.

As for when open source systems will appear that match such emerging standards, the first truly credible system was released almost a year ago. It's called Eucalyptus and it's backed by a commercial company.

As for which distribution would make the first move. Well Ubuntu Server Edition has included Eucalyptus since 9.04. Building a private cloud using open source technology that matches the EC2 API is almost as easy as apt-get install and it's going to get easier. We call this concept Ubuntu Enterprise Cloud and you can find details here.

It's also worth noting that Ubuntu Server Edition is also provided as official images on both UEC and on Amazon EC2, so you can run the same base image in both environments. Matching standards and using uniform images brings us a step closer to portability between environments but most importantly simplifies the process of bursting. It takes us a step further away from the dangers of the cloud net neutrality style argument that I highlighted at E-Tech'07.

It's also essential to build ecosystems around open source cloud computing which is why we work with companies like RightScale and CohesiveFT as well as our own tools like Landscape. Everything we do is around openness and freedom in the cloud computing space. This is not some ideological pursuit but simply a realisation that the future cloud markets will depend upon such openness to form. There is plenty of revenue opportunity without the need to tie people down in an old product mentality.
In short :-

  • You can already build clouds with open source technology matching the emerging standard of EC2.
  • You can already use single images across both private and public environments.
  • The technology is entirely open sourced technology and there exists no lock-in to a proprietary framework or solution.
  • It's free.
  • It's supported.
  • It's already in a distribution, go check out Ubuntu Server Edition.
  • You can already use a range of different management tools.

Ubuntu already dominates the linux desktop market, and from the reports I've seen recently we're going great guns on the server market as well. As far as I'm aware, we're the only distribution which provides you with a simple means of creating an open source private cloud and images spanning both private and public environments. Maybe we should get a bit better at shouting about it.

Well, I'm going to be speaking at Velocity and then I have a keynote at OSCON. I was thinking of a tag line for what we've being doing and a friend of mine, Alexis Richardson, chipped in with the following :- "Ubuntu is Cloud for Human Beings"


Wednesday, June 03, 2009

Double, double toil and trouble ...

It seems that the witches of Westminster have been caught with their hands in the cauldron. To avoid being burnt at the stake they've been resigning from the coven in an unnatural haste.

In some warped plot twist that would make Shakespeare squirm these blackened souls have decided that they have been wronged.

"Fair is foul, and foul is fair" these harpies cry as they organise the hapless master's doom.

You just couldn't make this stuff up. Are there no depths that MP's won't stoop to? Have they no shame? Not even a tiny sliver?

"Wool of bat, and tongue of dog, Adder's fork, and blind-worm's sting" - an MP's expense claim or a list of new job titles for them?

Tuesday, June 02, 2009

Making a big splash ...

All the buzz has been about wave recently.

The really smart bit can be gleamed from their statement to "help potential wave providers get started, our plan is to release an open source, production-quality, reference implementation of the Google Wave client and server, as well as provide an open federation endpoint by the time users start getting access.".

Oh, why oh why didn't they do that with Google AppEngine?