Network Optimization: how to keep your network in shape

NetOps (here we’ll be using this abbreviation simply to address networks’ team members) appear under a kind of double pressure from their clients or bosses: network performance should go hypersonic while costs should not just be kept low. Some near future idealistic calculations foresee network optimization expenses resolutely striving to 0.

“Do more with less”, we hear it a lot. Is mission possible, at all?

Something in the air

It is easy to compare networking to air traffic. Even some terms, like congestion and hub, sound and mean the same. The popular belief that an air crash hurts much more than a network collapse is only partially true: yes, everybody is alive after a network incident, but immediate losses, together with prospective losses due to operational instability, confidentiality breaches, etc. can actually surpass the cost of a lost aircraft. If this sounds too straightforward (and we certainly value peoples’ lives above all), vision network fail as a massive delay of flights caused by purely internal issues: inadequate plane maintenance, tower or luggage equipment failure, personnel strike, etc.

Aviation industry learned its lessons the hard way, but made right decisions. Everything there is precisely standardized and logically divided. Responsibility zones virtually do not interfere: ground operations, then runway(s), then approach, later area control – then all the same in reverse order for a destination point. Communication uses very specific and highly uniformed lingo. There is simply too little place for a mistake, even in the skies so congested prior to pandemic outburst. To get airborne and back safely you only need to follow the protocols – again, thoroughly detailed and globally accepted.

Things are not that bright (better to use “not so clear”) with networking.

Growing, not collapsing

The single yet sufficient reason to keep your network in shape is: the traffic is growing and will keep doing so. No need for a deep analysis of data explosion, at least not here & now. To name just a few phenomena generating loads of new traffic: IoT, streaming of all kinds, or simply Internet connection available in previously offline parts of the Third World.

Some numbers are truly fascinating: world traffic tripled from the year 2015 to this day, with annual 22% growth. How about Gigabyte equivalent of all movies ever made, that now crosses the Internet every 2 minutes? So, the Internet Superhighway itself is doing reasonably fine. Difficulties occur exactly the way they are on public roads: at the intersections, on detours, or when entering private estates.

The perfect network should be:

• Up, i.e. available for as much time as only possible. “Always” word may sound too idealistic, but in reality it’s not: currently accepted indicator of suitable availability reads 99.95%. The exemplar one is “nine nines”, 99.9999999% of time (or merely milliseconds of downtime annually). This includes all network resources and equipment.
• Low-, preferably very low-latency and completely jitter-free. The former is quite self-explanatory: it measures delay between data sending and reception. We are talking milliseconds again, but even these tiny slots can cause serious issues in, say, IP telephony or video conferencing. Jitter in networking means uneven, interrupt data flow: data packets are delivered not in the strict order; some are taking longer to reach their destination. This results, for example, in dropouts of syllables or whole words in calls or frames in video at the receiver’s end.
• Designed to withstand peak loads: not just having an adequate surplus capacity, but being able to properly report those peaks to NetOps and, in best cases, to offer or to automatically implement solutions to distribute the load and ease the situation.

We cited only the core requirements of an optimal network, in fact there are more of smaller ones. But the main issue that impacts networks’ performance lies deeper – so it is harder to see it (in many senses) and, respectively, to identify and overcome.

Aiming strategic targets. Uneasily

Unlike aviation, networks use variety of protocols and methods. To add to what’s already complicated: network solutions differ in time, so some modern and outdated components can be used within the same system, not matching each other well. This is usually a “double trouble” for NetOps. On everyday level these specialists are simply too busy: troubleshooting, patching bugs, etc. Yet when they devote some of their time to observe their networks “as a whole thing”, strategically – components’ mishmash often bars even an eager eye from seeing a structure end-to-end.

Business raised its demands for networks, so NetOps raised their concerns for networks’ performance – and therefore for optimization. The transparency issue is indeed serious: no less than 1/3 of network professionals indicated systems’ blind spots, insufficient end-to-end monitoring and difficulties in troubleshooting as their top worries.

Stingray Gateway a.k.a. The Optimizer

Stingray Service Gateway is designed with Deep Packet Inspection (DPI) process in mind. The platform inspects the designated traffic – with an instant filtering option, and may prioritize one kind of data over another. This leads to smarter, less stressed and more flexible bandwidth exploitation. Which, in turn, eases the total network load, eliminating possible negative traits: congestion, lagging, potential downtime, etc.

Stingray SG is particularly fine in monitoring all and any components of a network. Full, or summarized, or directional and AS Netflow reports; RTT with troubleshooting suggestions available thru API; click stream and marketing/redirecting options – all put together to evaluate the Quality of Experience (QoE) for a single user. Based on this information, the platform can initiate quality of service improvements. Needless to say, all Stingray SG monitoring is available on the spot, fully adjustable for the administrators’ needs and is no hassle to use.