01-11-10 | Blog Post
Cloud computing is a type of shared computing where one utilizes a hosting provider’s large-scale computing infrastructure. In other words, a provider with thousands of servers interlinked together, will rent out their massive computing capabilities to customers looking to only pay for the computing that they use.
One of the benefits of cloud computing is the flexibility it provides. In theory, you can use 1 unit (CPU, storage, RAM or data transfer) one second and 100s of units the next second and you only pay for the units you use with no minimum or fixed fee. There is no commitment and you can stop using/paying whenever you want.
The only investment and commitment you must make is writing your software to utilize the cloud providers computing architecture. Each time you switch cloud providers you have to re-write your software to match the cloud provider’s application programming interface (API).
The three enticing promises of cloud computing are:
Cloud providers have two disadvantages that must be overcome to enable the three benefits above for just about ALL computing. In essence, until they solve these problems, adoption will be narrow. Once they solve these challenges, we believe cloud computing will be ubiquitous.
The two primary issues are:
The SLA’s offered with cloud computing are lacking in substantial guarantees or assurances. Some of this has to do with the immaturity of the cloud computing model as the providers still figure out how they can deliver performance guarantees in a shared environment that can be impacted by macro events.
Internet and computing usage spikes on Black Monday or catastrophic news events are unpredictable – and planning enough computing overhead for these non-normal events can be costly for cloud vendors.
We believe, this SLA issue can be solved by technology. Using sophisticated monitoring and metering, and automated price bidding for computing cycles in heavy demand periods, cloud providers will be able to manage the allocation of real-time compute resources and throttle back second tier customers to deliver against a SLA for top tier users.
The second challenge is much tougher. For most businesses, cloud computing is too new and untried to risk their business life on it. It reminds me of when the Internet first appeared. People were very nervous about putting their data “on the net” or providing personal information on a web page. Now you can’t live without it (ever booked an airline ticket?).
Once the SLA problem is addressed, the benefits may outweigh this trust issue but until then adoption will be very sparse. And then again maybe the Cloud will never overcome the trust issue.
We believe the trust issue will be resolved. To us it’s no different than shared mainframe computing of the past, where obviously significant data privacy and ownership issues were solved.
To accelerate the adoption of the cloud in corporate computing we need to overcome (or bridge) the trust issue. The colocation operator is the answer to this challenge. Not only that but colocation operators can provide a very powerful and cost-effective channel to accelerate adoption.
Why do people colocate with their local managed data center operator, rather than go with the cloud, if it’s an option?
So, the question becomes how can the cloud leverage the trust delivered by colocation operators and use colocation as a channel to rapidly grow the market?
We think the answer is to develop an appliance that colocation operators can sell and manage for their clients. One that provides a certain amount of “umph” (CPU, memory, storage, # of servers etc.). The appliance would have the cloud vendors API to easily run cloud designed applications and could be designed with built-in metering that allows the client to only pay for what he uses.
But the real opportunity is the capability to allow an application running in the colocated appliance to “spill over” into the cloud by creating new servers and using resources in the cloud, outside the colocation data center.
Imagine if you will – an Amazon cloud appliance. A Software-as-a-Service vendor can design their prototype on Amazon’s cloud to Amazon’s API. When the time comes to production, the vendor can chose to continue running in the cloud, or move to a colocated cloud appliance if the location and security of their data becomes a trust issue for their clients, such as SOX, HIPAA or PCI requirements. Should the demand exceed their computing capability, they could make the call on which part, if any, they want to source to the Cloud.
It’s at that point that the client will really get the value of expandability. If they’ve had experience with the appliance and with the colocation operator, that can bridge the trust gap. Now they are in a private cloud and the colocation partner is there do to the support and provider services that mainstream applications and users demand at this early stage of cloud computing.
Some view the Cloud as a threat to colocation operators’ business model.
We don’t see it that way. From our perspective at Online Tech, colocation is a service business that adapts to provide an expanding range of services and solutions – by managing hardware, network and software infrastructures on a scale with a level of responsiveness and support that is difficult for most companies to do themselves. The cloud is just another evolution in the process.
The real question is how can the cloud leverage the colocation model to accelerate the cloud market?