Around ten years ago the ASP model was being touted as the next great wave of technology. The ASP –Application Service Provider model was trying to do Software as a Service (SaaS), and it didn’t ever really live up to the hype. So here we are ten years later, what has changed? Technology has grown up, and hardware is dramatically more powerful than it was ten years ago. The ASP models simply didn’t scale or make economic sense. We now have improved bandwidth, more powerful hardware, and the “Secret Sauce”-Server
Virtualization. Virtualization means we can slice up this hugely powerful hardware into digestible chunks and sell it off whilst still making a profit. The concept of multi-tenancy (more than one company sharing the same tin) was not achievable without Server Virtualization, multi-tenancy is key to the large cloud models.
There is a lot of noise around Cloud - this as a Service, that as a Service even X as a Service (XaaS)! What does this all mean, and how is this so different from existing outsourcing models?
The National Institute of Standards and Technology (NIST) have released a set of
documents around the topic of
Cloud Computing. I waded through them and decided to put an interpretation around them.
NIST breaks Cloud Computing down into
· Five essential characteristics
· Three service models
· Four deployment models
Wikipedia breaks it down into nine “key features” and the same service and deployment models, so what cloud really requires to define it as cloud looks like is still solidifying, but at least there seems to be consensus on the service and deployment models. Additionally the “essential characteristics” mentioned here are largely part of one or more of the “key features”.
Essential Characteristics
The Essential Characteristics of Cloud Computing are those things that are required of a service to make it qualify as true “Cloud Computing”. In other words, if it doesn’t do this, it isn’t Cloud Computing.
1. On-demand self-service
The ability for the client to self provision resources for themselves – storage or server resources
2. Broad network access
Here we are talking about the proliferation of devices at the end point that can receive the service.
3. Resource pooling
Shared access to computing power spread across multiple geographic locations –multi-tenancy.
4. Rapid elasticity
Scale up or back on demand. Burst capability.
5. Measured Service
Resources provided as controlled metered service to the client.
Three service models
1. Software as a Service
The provider provides access to the applications they are hosting.
2. Platform as a Service
The provider presents an Operating System layer and possibly an application toolset; the customer can then add customisations and or applications.
3. Infrastructure as a Service (SaaS)
The provider presents low level resources to the client, who can then build out operating systems and application layers.
{Desktop as a Service is a subset of IaaS}
Four deployment models
The four deployment models relate to where the services sit with relation to the customer and the levels of Customer – Community – Public clouds relate to physical and security boundaries.
By adopting common standards, loads could be ported between Clouds. Bear in mind the concept of “bursting”, analogous to bursting in network QOS terms, in periods of high load I could move some of my load to a Community or Public cloud.
1. Private cloud
The customer or provider creates cloud infrastructure for the private use of the customer.
2. Community cloud
Similar to a Private Cloud, the Community model creates a cloud for use by that community only.
This could for example be applicable to multiple subsidiaries sharing some cloud services.
3. Public cloud
The customer accesses cloud services that are made broadly available by the provider.
4. Hybrid cloud
A Hybrid cloud comprises two or more of the other models. Commonality between the clouds means that loads can be moved and or spread between clouds.
Conclusions
After studying the “Essential Characteristics” I realised that a lot of technology being sold as “Cloud” isn’t really “Cloud” – a lot of things are components of “Cloud” but “Cloud”, according to these requirements, is a lot more than “stuff you can get over the Internet”. This is a very high level outline to the topic of Cloud computing, but if Gartner/IDC etc. are to be believed, then Cloud is looming large on the horizon and we would all do well to have some basic idea of what it is.
Article Tags