Priorities in buying an IaaS platform
22nd Jan 2013 | 08:00
The factors in purchasing infrastructure as a service
Businesses have been exploiting the commercial benefits of SaaS (software as a service) for several years, and many are now asking themselves whether the principle can also be applied to their infrastructure needs.
Infrastructure as a service (IaaS) seems a natural progression if your business has been benefiting from the cost savings and efficiency gains that SaaS can deliver. Adopting IaaS platforms could provide significant benefits, but with an ever-growing number of providers, it's difficult to know which will be the best choice of service provider.
The current leaders in the IaaS market include Google, IBM, Amazon Web Services, GoGrid, NaviSite and OpSource. What all of these vendors have in common is that they are exploiting the power and flexibility that virtual machines now offer. In the world of IaaS, there are several processor cores available and no limits to storage space or computing power.
Care must be taken when comparing like-with-like, as some of the current vendors use their own terminology to describe processing power. For instance, Amazon uses "elastic compute unit" while Google uses "compute engine unit". IT managers should ensure they are clear about these definitions.
On demand infrastructure
The attraction of IaaS is clear: IT managers can buy just the server space they need, when they need it, all controlled via a web based dashboard. This self-service approach enables businesses to move away from fixed server platforms and their inherent refresh cycles and simply expand the number of virtual machines as and when needed.
The cloud has shown business that an a al carte approach to IT specifications is now relatively simple to set up and manage, and can produce substantial cost savings.
However, this has to be carefully assessed. There is a forecasting tool available, PlanForCloud, that can provide support, and there are a number of steps that potential buyers should take in making the decision.
For a start, it is important to look closely at the technical specifications of the virtual machines on your shortlist to ensure you are comparing like-with-like.
Also, the type of data storage that will be needed has a major impact on costs. In most instances, IaaS providers will offer local storage that is linked to the virtual machine and block storage that is billed separately, so there should be some consideration as to how block storage is set up for disaster recovery.
It is also important to evaluate what kind of applications will run on the virtual machine. If these are not processor-intensive but require a large dataset, businesses can chose a virtual machine with limited processing power, and spend the money saved on more storage capacity.
Operating systems in the cloud have also proliferated, with Windows, Linux and Unbuntu becoming familiar. Sticking with familiar operating systems will help IT managers to get new virtual machines up and running very quickly.
Integrated virtual machines that include some software installations have also become popular because of their ease of use. Businesses with older legacy applications that impact on their mission critical systems may be better served by standard IaaS installations that IT managers can then modify for their needs.
Security concerns have dogged cloud services for some years, but recently these fears have largely been allayed. IaaS installations have high levels of security including, for instance, compliance with PCI. Indeed, the Cloud Security Alliance has been taking major steps in this area to deliver secure platforms within today's cloud environments.
Moving much of the data management, storage and manipulation of information to the cloud is a clear trend, but one that should be approached with care. IT managers need to ask themselves a number of specific questions before making their buying decision:
•Is there a clearly defined application for IaaS? It's important to know what this platform will be used for. Data storage, data analysis and application development are just some of the potential uses for IaaS deployments.
•How much does your business want to pay for these services? Pricing across vendors can be confusing as it is difficult to compare like-with-like. Can you use less server processing power to pay for more storage?
•Have you considered the bandwidth that will be needed? The efficient deployment of IaaS will need robust bandwidth. Vendors will pass this cost on to their customers, so check what these costs look like before buying.
•Is your service level agreement (SLA) clearly defined? Uptime is a key component of the SLA, but also look at maintenance and any other costs that may not be immediately apparent.
•Can your vendor's platform scale on the fly? One of the key advantages of using IaaS is the elastic nature of the virtual machines. Look for how easy the platform allows for more server space to be added with auto-scaling features.
•Is Hadoop/MapReduce support important? Performing queries on massive datasets will mean implementing Hadoop. Ask how your vendor handles this and what levels of technical involvement your business will need to gain meaningful results.
•How much storage do you need and of what type? The levels of latency that your business can allow will determine what kind of storage will have to be attached to the IaaS platform. Block storage can get expensive, so check how much is needed and what this will cost before buying.
At the moment Amazon's Web Services reign supreme in this marketplace, but other pretenders to their throne are fast developing their own platforms. Google is a clear contender and HP, IBM, Dell, Verizon and Microsoft cannot be ignored.
From an IT manager's perspective this additional choice gives them many more options than had previously been available.