To start let's break down the tech speak. What are we talking about when we talk about Colocation in an IT Infrastructure sense?
Colocation is pretty simple, it’s where you house your own equipment in a data center that is operated by another provider (like an Internet Service Provider). Essentially you move some of your IT infrastructure resources into a datacenter and you then look after the administration of the hardware, software and network resources. There are a number of benefits to this solution, these include:
Uptime - Greater power redundancy is afforded due to the nature of data center power infrastructure. Additionally tightly controlled cooling also means your server and networking equipment should have a much longer and less fault prone life.
Connectivity - Generally it is easier and cheaper to get a high connection speed in a data center than it would be to have the same connection onsite.
Security - Systems are generally more secure from theft or physical intervention due to the security in place at these locations.
A quick google search of this this articles title will yield a number of detailed and conflicting articles that attempt to break down the cost of Colocation vs an Amazon Web Service (AWS) solution. Most of these articles come to the conclusion that colocation is cheaper than an AWS based solution. For traditional workloads, for large companies, this is generally true. However most of these articles generally gloss over much of the hidden costs of Colocation, and also fail to properly compare like for like features.
When doing a comparison of the solution most people just add the cost of the servers, rackspace, networking equipment and utility fees of a data center and do some basic math to work out an annual figure. But this isn’t really a good comparison when we look at the service AWS offers.
So what are the advantages of AWS over Colocation:
No hidden costs
Plethora of innovative tools and services that can be leveraged to transform the way IT works.
Let's break down AWS’s high level building blocks a little:
Each AWS region has at least three availability zones (AZ), two publically available and one hidden to perform replication and redundancy for many of their services.
Each AZ is part of your own Virtual Private Cloud network.
Each AZ is linked by high bandwidth, low-latency, redundant path links.
Each AZ is geographically separated from the others and is able to operate completely independently in the event of a failure.
When we starting looking at the level of redundancy and availability engineered in this infrastructure you will quickly need to start tripling some of the costs people are putting out there. Also most of these comparisons fail to factor in a number of the feature rich, fault tolerant services that AWS offers for businesses to leverage to think outside the traditional IT infrastructure box.
A great offering from AWS is S3. S3 is a truly redundant, highly available, highly scalable, feature rich storage service. S3 can and has been leveraged for all sorts of great things. In traditional workloads (virtual machines VM’s) you can use S3 to store snapshot data, leveraging the redundancy it offers to allow rapid restoration of VM’s to another AZ should one location be lost.
Another great offering is AWS’s RDS service. RDS is a relational database service which is managed by AWS. AWS will even look after the redundancy of the service. To engineer a similarly redundant service in-house would not only take a rather pricey salary, but also increased facilities management, power and utility overheads.
AWS offers a plethora of products that can be leveraged in a massive number of ways and they are continually innovating, improving existing products and releasing new ones.
Another additional consideration is lifecycle management. If you are managing physical infrastructure you need to budget and plan for a refresh of the equipment at least every five years. The cost of this to the business is far greater than just the cost of hardware. There is also the staff to perform the refresh and not to mention the interruption to general business and management overhead. With AWS this isn’t required at all. All you need to worry about is the lifecycle of the application.
In summary the value of AWS should not be quantified in just comparing server and storage costs from a single colocation facility. AWS’s service and offering make building a comparable solution in house an extremely time consuming and expensive endeavour. While Colocation can be a viable option for some businesses with traditional workloads; for those who want to focus their energy on what their businesses do best, AWS is definitely a great solution that offers many opportunities to transform the way your IT works for you.
By Rhys Harris, Senior Systems Engineer, Myrtec