When something is a very precious asset (like network bandwidth), first know what is happening there - “You can’t manage what you can’t measure”
Manoj Narang (founder & CEO of Tradeworx ) says that “Nobody’s making extra money because of them (networks); they are a net expense to the company” – (ET, p.14 31 march 2012).
Very true – network is a huge expense to the company – something which is creating a huge hole in the expenses and will continue to grow exponentially as the internet, cloud, and the number of web applications continues to grow the way they are growing without mentioning the growth of the network data via mobile.
Many industries biggest complaint is that my network is not responding at the rate at which it should have responded or was responding so far which is affecting my productivity and worse increasing my frustration and guess what network manager has no clue either other than going back to the CIO or the CEO asking for more expensive bandwidth. Especially in Trading & financial firms where even milliseconds delay (or fast) matter and worth millions for them and they are ready to afford high speed bandwidths and they do. As per financial market research firm Tabb group, about 55% of the US equities trading volume come from firms using high frequency bandwidth.
Other 45% data generating companies or other industries or those SMBs who cannot afford those high end super-fast super big bandwidths need not sweat. There are other ways one can speed up their traffic at their end along with a combination of not-so-expensive MPLS/other networks. I will not talk about MPLS or other high speed networks but read on about how to make your side of the network not choke your side of the bandwidth, and some research suggested that 50% of the choke happens outside and other 50% happens inside. Let’s fix the inside part as it is easier and in our control.
When you need your business critical applications to use the precious expensive bandwidth, there are always many non-biz critical or mundane routine backups and replications hogging the bandwidth and they don’t know nor aware that they can actually slow down when a biz critical application data arrives and then move back to their normal speed when the biz critical ones are done – how can they – they are not intelligent packets :-) But yes you have technology and products (like i7 networks or few other companies across the globe) which can actually sit at the edge of the network (or as in i7 case – even on the cloud) and sniff the packets (deep enough to know what kind of traffic it is, even if it in encrypted), auto find out what kind of traffic it is and use self-intelligence to see whether it is critical or not and prioritize (or even contain if it is entertainment or stop if it is bad) the critical apps and critical data and get the max out of your precious expensive bandwidth – by this you almost won ½ the war :-)
Coming back to kind of products, new generation packet visualization and shaping or prioritizing products are in the making. They can now visualize, manage networks at user level, application level, devices level etc to such a granularity, one will be amazed to know what is happening in today’s network and that too all at real time giving enormous data and control to network monitoring team to catch issues at brown-out time itself rather than at black-out saving precious money both by preventing such outages as well as enhancing productivity. Even metrics and dash boarding of metrics has become very advanced with metrics being runtime, collect all data rather than just sampling, great insights into analytics with different way of seeing depending on who is consuming it (network engineer, to NOC center people to CIOs to CEOs), Application centric to device centric to user centric to biz hour data reporting with clear details on who/how/what the bandwidth being used. All possible because, it accomplishes that by looking (inspecting) at every packet that traverses the network, understand its contents enough (not complete and especially not data so that privacy is still breached), figure out where it originated, where it is destined to. Sometimes the packets are masked to cover up all these and that can found out out too as one doesn’t inspect just few packets but even the flow of packets reveals much more info. As they say you can’t manage what you can’t measure – step 1 – let’s see & measure what is happening on our bandwidth first :-)
(Author is the CEO of i7 Networks which works on network traffic visualizer and shaper. Done via on premise (either appliance based or so or on the cloud)