It’s tempting to wonder if computer industry analysts are guilty of changing their outlook to suit the circumstances whenever it’s convenient to do so. In the server sector of the market, that would certainly be easy to believe today when one considers the trend of Edge Computing.
Edge Computing has been referred to as outward scalability, which in simple terms means taking specific dedicated functions that would otherwise have been provided on a central server, and packaging them up on stand-alone hardware and software platforms to put them at the edge of the network.
At first glance, the concept of Edge Computing might seem mutually exclusive to another server industry trend, that of Server Based Computing. After all, we’ve just finished accepting that distributed desktop PCs were yesterday’s idea, and that server based computing is the future. We accept that the annual cost of ownership of business PC systems is vastly greater than the raw, capitalised purchase price when factors such as licensing, support, upgrades and general maintenance are included. We acknowledge the business sense of moving to a centralised model where one or more large servers can homogenously act as a single point of management for support of a network of thin clients or network PCs.
But if we are planning our business server strategy based upon fewer, larger servers that can accommodate multiple tasks and applications from the centre of the network, then how can it make sense to be planning more, smaller servers at the edge of the network?
The reality is that these two concepts need not be mutually exclusive. Techno geeks who talk about servers refer to the term scale, by which they mean the ability of a server to grow and provide additional resources as application demands expand. But there are two ways to scale. One is obviously upwards, by adding more power and capacity to a centralised server, but the alternative is outwards, by offloading specific functionality from the central server to free up resources, and then adding a dedicated server to perform that function independently.
The concepts of upward and outward scalability are complementary in the sense that they achieve the same end result; more server resources. Moreover, rather than complicating the management process and adding to costs, the migration of servers to the edge of the network actually helps simplify management and reduce cost.
The reasoning behind simplification lies in the nature of the applications and servers that are typically scaled outwards. Instead of the term server, devices at the edge of the network might more properly be termed appliances, since the dedicated nature of their function implies a device that is simply turned on or off. There is little to manage or administer, and expensive central servers have their costly resources freed by replacement with an inexpensive appliance that works as easily as a toaster.
The types of appliances used in Edge Computing are really stripped down servers, with limited hardware and software that are designed to do a specific task. Optimised for specific requirements, typical functionality might include web caching, content delivery or media streaming. A complex example might be an e-commerce application where the every day demands of general enquires to a web site might be easily handled by dedicated appliances at the edge of the network, but where complex application processing for catalogue lookup, ordering and credit card processing are handled by servers at the centre of the network.
The concept applies equally well to more mundane activities, and a typical financial institution in Nassau might readily have a domain or email servers at the network’s edge, freeing centralised, costly applications servers from those routine, everyday tasks.
The reality is that in today’s economic climate, where all IT Directors are expected to get the best bang for jealously guarded bucks, a strategy that incorporates centralised server concepts with the use of dedicated network appliances simply makes sense. The trick, of course, is being able to properly recognise what goes where to build a network that is cost optimised.
Paul Hutton-Ashkenny is the president of Systems Resource Group Ltd. and Bahamas On-Line. Views and opinions expressed do not necessarily reflect those of SRG or BOL. Questions/comments to: P.O. Box N-3920 Nassau, Bahamas or e-mailed to: info@srg.com.bs.
