Tuesday, March 27, 2012

How To Deal With Sudden Massive Bandwidth Demands

You’ve been cruising along with fairly stable but slowly increasing network bandwidth demands for years. Then, one day, you are faced with an “out of the blue” requirement to crank up your WAN bandwidth by a factor of 10x or 100x. How did that happen and what can you possibly do to deal with this kind of uncertainty?

Ways to ease the mad scramble to increase WAN bandwidth...It would be strange, indeed, if everybody in the company suddenly decided to start downloading HD movies on the sly. Would it blow your bandwidth allocation? You bet. Does it make any sense? No, of course not. Bandwidth demand doesn’t suddenly mushroom for no reason at all. If you get caught unable to support the need, it’s likely that there is a logical reason and one that makes a lot of sense in retrospect. The question is should you be able to see this coming in time to plan for it?

Sometimes you can’t help but get caught off guard. At a large company where I was working I got an email message with a curious attachment. When I opened it, there was a short video animation of a dancing baby. Not having seen this before, I watched it a few times and sent it on to some colleagues. Turns out that this was an early example of viral video. Everybody in the company was getting this video and passing it on to everybody else.

The IT department was flummoxed. All of a sudden they were running out of disk space. It took a day or two for them to figure out that all of our mail boxes were full of separate copies of the dancing baby video. They were multiplying faster than baby rabbits. Soon a missive came out from IT management demanding that we stop sending this thing around and delete the copies we already had. Crisis averted. Gigantic new disk drive order cancelled.

Today we grab everything off the Internet. Instead of getting a file in our email, it’s a Web link. You probably won’t be filling up the storage drive farm anymore, but any content brought down to the screen is going to have to traverse the WAN. It doesn’t take much of a viral anything to clog up a dedicated Internet connection if the file is big enough and enough people want it simultaneously.

Here’s a more recent example of where IT departments are getting caught short. C-level management gets a case of cloud fever when they see the potential cost savings and insist on moving every application possible to the cloud as fast as possible. In the mad scramble that ensues, the local data center goes dark while the same apps light up at a remote cloud provider. With nearly infinite CPU cycles and GB of storage on demand, everyone expects that the days of bottlenecks are over and that every process will just scream. Shock, horror, and disappointment permeate the office when users discover that formerly snappy responses from the system have gotten sluggish.

How can this possibly happen? Adjusting performance at the cloud doesn’t seem to make any difference. What got forgotten is that all that traffic that ran smoothly on the LAN all of a sudden transferred to the WAN. The line capacity is too small and the latency is too great. Whoever had the bright idea of saving a fortune by accessing the cloud via the Internet is finding out that network neutrality favors Netflix viewers as much as Salesforce.com users, as everyone’s packets take their turn on the Information Superhighway.

This scenario can be avoided with carefully planning and enough lead time to get the right bandwidth resources in place. That can be a challenge when bandwidth provisioning time is the long pole in the tent. You can hedge against future upsets by installing connections that have easier scaling than traditional telecom services. Carrier Ethernet is just such a service. Install a 1 GB port but only commit to 250 Mbps of service right now. When, out of the blue, some VP gets every employee signed up for an online video training course on the latest trendy productivity or quality initiative, at least you can grab the phone and tell your provider to crank up the bandwidth immediately.

You may also be served better by contracting with an MPLS network provider instead of running your own private network of point to point dedicated lines. It’s a major reconstruction effort to upgrade bandwidth on all those lines, especially if you have to go from copper to fiber. Your MPLS provider may be able to give you higher committed rates quickly or allow you to burst to handle unexpected peaks. Best to discuss these possiblities before signing a service contract.

It’s always possible to get deluged with a difficult or impossible to foresee spike in resource demand, including bandwidth. Know that there are services available now that make scaling BW up and down a much easier and quicker process than traditional approaches. Do a little planning now and see what’s available in Carrier Ethernet and MPLS Network services designed for rapid scalability.

Click to check pricing and features or get support from a Telarus product specialist.

Follow Telexplainer on Twitter