IaaS or Infrastructure as a Service has been prominent since the time Amazon launched the EC2 or Elastic Compute offering. But, was there no IaaS before EC2?
Infrastructure in the computing world has defined everything starting from server hosting Data Centers to servers, storage, networking, cabling, cooling, etc. Infrastructure was the layer that enabled software to function and execute. The software layers included the OS and the applications that were hosted on it.
In the physical world, long before EC2 became available, third party DCs rented out space, power and cooling in their DC for organizations to host their own compute infrastructure. Typically, they charged a subscription cost for these services.
After EC2 came into prominence, Compute infrastructure, which always took a couple of weeks to become available after a request was made, was instantly available at a much lower cost. Compute infrastructure which was previously seen connected to physical infrastructure now had no linkage.
There has been a lot of innovation to perfect this model. New vendors like Rackspace, Terremark and Savvis provide features that include the ability to choose from a list of available “templates”, to configure the templates, etc.
The DCs that host the compute themselves are now able to provide almost 0% downtime, enhanced security that includes perimeter security along with biometrics, and other forms of access control.
Pricing has also seen innovation. Based on the time of day and the overall load, a user can avail a “spot” price. This keeps both the provider and the user happy as more of the infrastructure is being used and the spot price is lesser than the list price.
DCs also meet and advertise a set of compliance requirements such as HIPAA. Vendors like Rightscale and Firehost support such multiple certifications for industries like healthcare and finance.
All the above mentioned vendors represent the public cloud.
At the same time in the Enterprise arena, vendors such as VMware, Microsoft, and Eucalyptus want to create an Amazon like experience. A user can access a self-service portal, select templates and also expect to be charged in the same subscription based manner; all of this with the support of secure multi-tenancy.
I believe that this is the time for the next generation of IaaS. A few factors which I think would define the next generation would be:
The first factor would be the Infrastructure that would have “Application Intelligence”. If the Infrastructure layer can be referred to include everything starting from hardware up to the OS, and any run time libraries required by the Application, the Infrastructure should enable high performance delivery. The Infrastructure needs to be intelligent, learn from the behavior of the Application, learn how it is accessed during different times of the day, what the load conditions are, etc. It should be able to tweak the kind of resources provided to the different components, so that the overall Application performance never suffers.
The second factor would be where the Infrastructure “understands” the notion of cost, the cost of running the servers, the cost of software licenses, the cost of the power, cooling etc. With this understanding, the Infrastructure can minimize the cost element and, increase the ROI.
The third factor would be the concept of “seamless” infrastructure. While this is available in pockets in the form Amazon VPC, this can grow on to become a lot bigger. An organization can see its networks seamlessly extending or shrinking based on the need to add / relocate compute.
The fourth factor would be Infrastructure “consolidation”. SMBs will cease thinking of having their own infrastructure and rely totally on Public Data Centers. Like power, cooling, water etc., compute will also be a resource that can be metered and accessed any time.
The fifth factor would be “fault tolerant” Infrastructure. Fault tolerance or high availability is typically linked to servers or applications, and would need to extend to include the entire Infrastructure. So in the case of the primary infrastructure not being available, the secondary infrastructure would kick-in instantaneously.
Our roadmap addresses building intelligent Infrastructures that are Application and Cost aware and have the ability to provide “seamless” connectivity, “consolidated” and “fault tolerant” Infrastructures. Keep following us.