Being a Cloud evangelist from an early age, I have always known what cloud is, how to use it and what the fundamental benefits of it are. In the space that I have been in for the last two years I have constantly looked at what the feature sets are that drive businesses to the cloud. I have always had the conversation with businesses about lowering costs and centrally-managed systems, which save time and allow for business growth through reducing complexity and allowing focus in other areas.
For some businesses this exercise has resulted in costing more than the traditional architecture that they are used to. I have seen that a lot of the time this has been a result of trying to achieve the best scalable solution, as being able to scale has been one of the focus areas that allow a business to reduce costs.
With cloud always being marketed as scalable, I have seen that most hardware vendors market in a similar way, blade servers and blade storage arrays are also scalable as well as most hardware appliances ranging from networking devices to security appliances, but what is the difference in the two types of scalability?
So what is scalability all about?
When I did some research on linear scalability vs. the scale-out approach, it seemed that scalability is a very broad term on the internet with regards to the different types, and what scalability should allow for. The most common comparison that I came across was ‘Linear’ scaling vs ‘Horizontal’ scaling.
Linear scaling, in the simplest words, is the ability to scale up or down (like you learned in school with line graphs). Horizontal scaling is the ability to scale left or right.
When I looked deeper into this, linear scalability allows us to increase or decrease resources to a server, adding or reducing CPU, RAM and Storage.
Horizontal scaling allows us to add more servers to a cluster and split workloads or take on new workloads.
In my opinion, I believe that this thinking is still very focused on the underlying hardware infrastructure and architecture and in a software defined world, where does this truly benefit businesses?
I am able to add more resources to my server and I am also able to add more servers, but is this not just solving a capacity problem when it comes to over provisioning or under provisioning?
Early in the game, I believe that over provisioning was one of the biggest challenges that businesses faced, they would buy infrastructure to last the business 5 years and barely got to use 50% of it. This means that they paid for 50% of something they never got to use and at the end of the 5-year period, they needed to refresh the hardware as it would be outdated and not supported.
When we are thinking about infrastructure and architecture, there are so many approaches that how would a business know which one to use. The scalability of underlying hardware is always something that first comes to mind and in the true essence of cloud this should be the least of our worries.
In the next part of this series, I talk about the ‘Scale-out’ approach. Watch this space…
Matthew McEnroe is the Product Manager for Azure/Hosting at Tarsus On Demand.