Amazon’s announcement today on Virtual Private Clouds (VPC) and Werner Vogels’ enthusiastic blog response have been used by advocates to validate positions for and against the notion of the “private cloud.” We’ve been part of that debate ourselves (see Rise and Fall of the Private Cloud, Beware of the Wolf in Blue Clothing, and in a dialogue here with Christopher “Fire Hydrant” Hoff). While we strongly advocate technologies like virtualization, we’re critical of the notion that a private cloud can be at the center of a future state IT strategy.
Two takeaways from today’s announcement of Amazon’s Virtual Private Cloud:
- It’s very exciting. This is another major step in the right direction to seamlessly integrate with, but still isolate, legacy on-premise investments that are a drag on enterprise IT departments.
- Vogels’ own blog spells it out clearly – “Private Cloud is not the Cloud” – it’d hard to be any clearer. Amazon’s providing a Virtual Private Cloud is as much of an endorsement of the private cloud as their Virtual Private Storage (S3) is of on-premise storage.
One of the ironies of the public cloud vs. private cloud debate is that the private cloud isn’t really a cloud and the public cloud isn’t all public. Every “private cloud” to date has proved to be just an upgrade to your own data center – you still own the cost of all IT management. Alternately, every “public” cloud provider has aspects of being ‘private’ – it’s just a question of what level. Although you can access the “public” cloud almost ubiquitously, your data is yours alone and no one else can see it. What makes a cloud “public” is that the some aspect of the hardware, OS,database, app server, metadata, etc. is shared across customers. For Amazon VPC, nearly every layer of the stack is private, it just happens that the underlying hardware they sit on may be the same.
But even this misses the bigger story behind this week’s announcement– Cloud computing allows for a nearly unprecedented rate of innovation in a very significant part because of multi-tenancy; Amazon’s story is a proof-point of this effect. Some will note that Amazon competitors have offered similar features, but I don’t think there is anyone who a few years ago would have predicted Amazon, or anyone else would deliver this list of services so quickly to the entire market. Look at their timeline of advancements since their days of only being just an online bookstore:
- Simple Storage Service (S3)- March 13, 2006
- Simple Queuing Service (SQS) – July 11, 2006
- Elastic Cloud Computing (EC2) – August 24, 2006
- Flexible Payment Services (FPS) – August 2, 2007
- Simple DB – December 13, 2007
- DevPay – December 18, 2007
- Elastic Block Store (EBS) – August 20, 2008
- Cloudfront – November 18, 2008
- Elastic MapReduce – April 2, 2009
- Virtual Private Cloud – August 26, 2009
And these are just major new services that don’t include improvements to existing services (e.g. EC2 running windows).
Innovation and advancements at this pace is a testament to strong strategy and execution, but only possible by relying on the fundamental tenant of the cloud – a shared multi-tenant delivery model. In just three years Amazon has provided a set of capabilities to the entire market that most companies and IT shops would love to claim for themselves. Amazon successfully moves faster than the on-premise alternatives by constantly improving a single code-base instead of installing and managing independent versions of software deployed at each customer. This is the secret sauce of multi-tenancy and explains why certain providers can move faster than everyone else; you’re not fighting the gravity of multiple versions and codebases.