A look at scalability and Azure Part 2 – Scale-Out A look at scalability and Azure Part 2 – Scale-Out
TOD's Matthew McEnroe looks at Azure's ability to scale out in the second part of his cloud scaling examination. A look at scalability and Azure Part 2 – Scale-Out

When I did some research on the term ‘Scale-Out’, I found that this is very much in line with the horizontal scaling. Most pages that I looked at, had spoken about this in their own words but the underlying message was all leading back to the horizontal scaling approach. It was all based on adding nodes to existing architecture to increase performance and take on new workloads. This was all focused on how to get the best optimisation out of your hardware.

Personally, I believe that the term Scale-Out can be pushed beyond hardware and even software. The reason I say this is because with virtualisation technologies on the market allowing us to scale hardware and software, we can achieve the same results no matter which Cloud platform an organisation uses. Most cloud platforms anywhere in the world can afford you the ability to scale up your servers and increase and decrease your usage on a pay as you use model. This is nothing new and solves the problem of over or under provisioning, however this goes a step further.

Looking into Microsoft’s Azure offering I have learned a new meaning of Scale-Out architecture. Azure has taken the Scale-Out approach far more aggressively with redefining the way in which businesses are able to compute. As most organisations already know, on Azure you are able to scale up, down, left and right with your servers, but they have taken it a step further with their Platform as a Service (PaaS) model and Software as a Service (SaaS) model.

Traditionally, this model was more attractive to ISV’s and companies that natively built their own software and sold it to businesses as a service. This model hasn’t changed. It has just become better and has made it easier for organisations in every vertical to optimise efficiency through this model.

As a very basic example, if a business needs a SQL database for a production workload or a dev test, they would have needed to build a server on premise or in the cloud, install SQL and use it in line with whatever application they needed. Azure takes all of that away with SQL as a service. There is no need to build a server or put an extra workload on an existing server – that is already done. The service allows you to use SQL on a per instance basis and allows you to leverage off a latest and fully updated version of SQL that you are able to use on the fly when you need it. When you are done with the service you simply stop the subscription and you will no longer be billed for it.

No need to build a new server – you are able to scale your application on hardware you don’t own!

With global applications such as SAP, WordPress, SQL and many other software on the international market, businesses no longer need to build VM’s or physical servers to load these applications. They also reduce a much higher cost in time and in most cases no longer need the purchase of applications.

I don’t think that in the near future building VM’s will be irrelevant, but we can look at an approach to use legacy workloads on servers in the cloud and then leverage off the Software and platform as a service models to reduce the amount of machines that we need to manage.

Azure is cost effective and offers you the application scalability that you need. With having your Windows or Linux VM’s hosted with Azure, you are able to utilise many of the services on an application layer that will no longer require you to build more servers.

If I look at the backup and recovery space, the usual model servers would be to build a central backup server loaded with a management console. With Azure backup, you no longer need that central server because Azure already has a dashboard console for you to manage your backups that store the data. The same applies to many applications that a modern business will use.

In the next part of this series, I talk about the Machine Learning and Analytic Engines.

Matthew McEnroe is the Product Manager for Azure/Hosting at Tarsus On Demand

Leave a Reply