May 17, 2013
Let’s do a quick calculation. How many employees do you have, rounded up to the nearest 100? What percentage of them are mobile? Have they sourced their own devices through BYOD? Do you use VPN access through to a private data center?
Hang on… what has the last question got to do with all this?
Let’s assume that you are a middling-sized company with 1,000 employees. Let’s also assume, for the sake of argument, that 40% of these are “mobile” — as in, they use a device to carry out work activities while outside of the office. That’s 400 people who are making demands on the corporate ITC platform at some point. Finally, let’s assume that there is a 20% concurrency — therefore, 80 people are all working against your resources at the same time.
What is the WAN speed in and out of your data center? Again, let’s assume that you have a decent bank balance as a company and that you have gone for a pair of load balanced, 100Mb/s leased lines, so giving a total maximum capability where both lines are working of 200Mb/s. For those 80 people, they have the capability to access 2.5Mb/s if it were to be dedicated to them alone via that VPN. Sounds great, doesn’t it? Not quite what you are likely to be getting at home (in many cases, probably around a tenth), but good enough, surely?
Unfortunately, though, apart from the fact that the “last mile” for many of these mobile workers will be over a highly contended public link of WiFi, ADSL or 3G, there are 600 desk-bound people who will also be using those lines from inside the organization — and if you are using VoIP as well, then a whole raft of phone systems. Some of these services may be using virtual connections with dedicated bandwidth — so taking that resource away from any real use even when the service the link is there for is not in use.
As you can see, the amount of bandwidth available per employee or per task starts to fade away quite rapidly. Taking a 10,000 employee organization with a single data center, maybe using a degree of video conferencing alongside everything else, starts to bring the available bandwidth per workload down to what an individual would have expected back in the 20th century — and yet they believe that they are receiving 21st century, superfast speeds.
You could just throw more money at the problem: use 1Gb/s lines or build more data centers to share the load. You could use external, co-location data centers and ensure that the owners have enough bandwidth plugged in. And yet, as the inexorable march of growth in IP traffic grows, you’ll still run up against a brick wall of bandwidth constraints at some point.
WAN acceleration can really help — instead of every packet transferring from the access device to the service platform, “static” content (i.e. that which doesn’t change much) can be cached either at a server closer to the access device or on the access device itself. The inherent “chattiness” of certain traffic can be curtailed, with only the “real” IP needs traversing the WAN. Packets can be re-shaped to be more efficient in how data fills the pipe. Data can be deduplicated to remove the transport of the same information down the same line more than once.
A modern organization needs good WAN connectivity, and its mobile workforce will be increasingly demanding that this is in place. Unmanaged use of what can be seen as very fast connections will lead to a poor end-user experience and lots of complaints through to the help desk. Get to grips with the problem as early as you can — monitor usage, manage bandwidth using acceleration — and ensure that everyone gets the best that your connections can truly offer.
Image credit: brianac37 (flickr)