« September 2009 | Main | November 2009 »
Only a short post this week, to bring to your attention a new white paper we wrote about Wan governance. Its access is free, so do not hesitate to download it here. French and German releases are also available (or will be soon). I'll be very interested by your comments and attentive to any feedback and suggestion you would accept to provide. Do not hesitate to enter in the discussion!
.jpg)
Such KPIs are usually composite objects that consolidate a large number of technical metrics (like delays, jitter and loss, response time, bandwidth, amount of data, etc.), compare them with thresholds (e.g. minimum and objectives) and finally normalize them so they can fit in a predetermined scale. Application Quality Score (AQS) is one example of such simple indicator that may be represented in many different ways, from list of values to colors in multi-dimensional graphical maps, and that anyone can intuitively understand.
Voynich manuscript may be meaningful (although no one really knows) but is by no mean an easy to understand document. Hoax? Exotic language? Glossolalia? Why not Roger Bacon's secret writing? Attempting to decipher it would be an interesting hobby for "Da Vinci code" amateurs, but who would like such a challenge with Wan Governance KPIs?
Illustrations: extracts from the entire manuscript on http://commons.wikimedia.org/wiki/Voynich_manuscript
Maryanne Wolf (1) uses this image to illustrate the complex transformation from signs to letters to sounds to concept, that happen in the reading brain. C, A, T are three independent signs we hard-wired our brain with as the first step of learning to read, and unconsciously decode as 'letters'; each of them does not mean anything 'per se'. When we see them altogether (CAT), a decent English reader can associate the three sounds /k/, /a/, and /t/. Finally the coupling of these sounds gives birth to the conscious concept of
.jpg)
Similarly, WAN governance is made of pieces that, taken separately, have totally different meaning. It starts from techniques like pattern matching, shaping, tcp ack anticipation, CIFS decoding, object caching, etc… They are the signs that, when properly decoded, become coherent 'technical features':
In their turn, these 'technical features' are the 'letters' that, when linked together in the right manner (ideally by Autonomic networking) will be able to collectively deliver higher level users benefits like:
Finally, the association of these benefits within a coherent process and methodology will make appear the conscious Wan governance concept and the corresponding alignment of the network to the enterprise business and IT strategy.
Of course, one can say that all of this is far-fetched: after all, kids & cats understand each other well before being able to read, without any need of conscious steps and rational explanations that heart blinded adults need to understand their world…
(1) From Maryanne Wolf's "Proust and the Squid: the Story and Science of the Reading Brain" (again…)
An old one, broken, powerless, abandoned, forgotten in the bottom of a drawer. Only this instrument can provide you, at 100% certainty, with the absolute accuracy that even scientists do not dare to dream about. Twice a day: once in the morning, once in the evening. Easy. Cheap.
This is what we got with legacy policy-based networking where people needed to precisely define the expected traffic conditions:
… then to decide and apply static rules supposed to handle efficiently this traffic at each network node or access (limiting, shaping, priorizing, queuing, coloring…). At best, those rules will apply twice a day, when the traffic gracefully accepts to match with the model. Of course, traffic probably never looks like expected, so these rules might well be never relevant… Cling to hope…
Reality is much more dynamic and complex. Whatever the scale you look to the traffic, you'll see ups and downs, hills and valleys that correspond to traffic changes: peak time of the day, peak time of the hour, peak time of the minute… Someone starts a transaction, her colleague set-up a call, a third one downloads emails while looking to a video on the enterprise tv channel: where are the correlations? How can you forecast?
There is a mean to make the network clock up and running: Autonomic networking. With A.N., the traffic – I mean real traffic, not guesses - is understood in real-time. Self-learning and self-adapting, Autonomic networking control closed-loop dynamically set-up and refines traffic policies: network and traffic always match, whatever the conditions. Usage is optimal, efficiency is maximal, Wan Governance is global. Anytime.
Ok, Autonomic networks may be 1 second late or 0.5 second in advance. But who care about plus or minus 1 or 2 seconds to get a train, be on time at a business meeting or even at a romantic rendez-vous?
