news / tech talk

I have written before about cloud computing and separately about Virtual Machines (VMs) but I would like to go over some old ground with new information and bring the two together. To summarize, cloud computing is simply using hosting, network, and storage services provided by someone else. The provider maintains the core infrastructure and the customer (you) maintains the operating systems and applications. Whether a hosting provider like Rackspace or a more abstracted service like Amazon or Azure, it’s still effectively the “cloud” to the customer.

If you’re not familiar, VMs are “guest operating systems” that operate from large software files on a shared hardware server. The hardware server operates a stripped down operating system  called a “host operating system” and is designed to run with little overhead and multiplex several VMs on a single hardware server. The point of virtualization is to take advantage of the fact that a single operating system and it’s applications often under-utilize the hardware they are sitting on. One can reasonably run several operating systems on a single set of hardware, especially with the large memory capacity and multiple CPU cores available on hardware today. Other advantages of VMs include their existence as a somewhat portable large file which means they can be saved or “snapped” at a certain point in time as well as their ability to be easily and exactly copied, and so on.

While many cloud services support both VMs and dedicated hardware operating systems, the trend is definitely towards virtualization. Something to consider in looking at these services is whether you need “fully managed” services or simply services. Fully managed services usually mean that you will get help setting up your systems, monitoring them, and maintaining them with security patches, updates, and so on. Fully managed services cost more than unmanaged services and are not offered by the really large cloud services like Microsoft’s Azure and Amazon’s S3 but they are offered by many traditional hosting companies like Datapipe, NetArx, RackSpace, etc. It seems expensive but when you factor in the cost of dedicated staff for the system and network administration, it can be a bargain.

The big cloud services like Amazon, GoGrid, and Azure are very much “self serve”.  If you are old enough to remember the difference between Full Service and Self Service gas stations, then you will know what I’m talking about. Self Serve means that a customer needs to have a knowledgeable IT staff who know about capacity planning, security, patching, and system administration. These services are best suited to a company that already has a knowledgeable IT staff and just need the ability to run their applications remotely in high quality data centers where power, storage, etc. are available. If one doesn’t have this staff and knowledge already, then these systems will just seem confusing unless one gets help from another party. If one has the right staff, then these services are excellent as they provide a number of tools to make many things simple and easy. More importantly, they provide a smaller company with access to high quality, super scalable infrastructure that they would otherwise never be able to afford.

The combination of cloud services and virtualization offers incredible flexibility for many companies, easing and simplifying a lot of burdensome tasks. For example, one can take a production VM and copy it exactly as a backup or to try adding some new functionality. If the functionality tests well, they can feel secure about making a change or simply shifting traffic to the new system and leaving the old one offline as a backup. Without virtualization and the near-limitless hardware capacity available in a large cloud service, doing what I describe above could still be done but it would be expensive, painstaking, and very slow to accomplish.

I am still sometimes annoyed by overblown “cloud and virtualization” talk as marketing jargon but if one knows what it is and what it does well, then like any tool it can be used very effectively.