Next step

Days just pass by, no work gets done. In office I surf and at home I entertain myself. In blogs I write things and never seem to get things done... Until yesterday :-) I found my touch. I was back among the books. (I hope the Indian batsmen get back among the runs too). In other words "phod diya". (college lingo). Downed a whole O' reily book on Perl in just 2 nights. Well now that it is done, I can concentrate on other stuff... And hopefully I can get my work done double quick in office, and focus on the real issue.
AVOIDING BUGS...


SMART-TESTER

lets say we want to simulate ospf neighborship between 2 routers.

starting this practise of writing down my ideas as soon as I get them...
1) Ospf protocol simulation using routers as objects with attributes like interfaces etc.
2) simulation "knows" what should happen and live routers will have to conform
3) this simulation will be superior and relatively bug free because :-
(i) It does not have millions of other features to support
(ii) It is not constrained by complex hardware issues etc.
(iii) It is a light weight application compared to the actual protocol
(iv) It will have model based testing brains ;-)
(v) In future we need to add "machine learning"



Thanks Pradeep Sindhu et al, Juniper routers can speak in xml
boy.. I like that so much.. after the dead text processing in Cisco...


hohahahaha
I feel so good since I stared writing down my thoughts.. It is conservation of my thought energy.. I am not allowing my it to dissipate. I have been guilty of that so often. Simply procrastinating does not get you anywhere. One has to move his ass. And I swear damn it, that is exactly what I am going to do.

Objective : To proofcheck a network. Not just test it.. but to prove that it will work the way it should. (million dollar SLA's are signed for this.. and maa company would be happy to spare a few thousands for me) This product has a lot of value.

Can we have a simulation of a network? with ALL the complexities built in it? no.. KISS: keep it simple, stupid. I would be better of designing seperate simulations for seperate features. Built on a basic simulation of routers and interfaces... OOP anyone? A classic and intelligent implementation of OOP is required. Rajaram, (my baap working in Amazon.. and by now I bet he is the baap of OOP too) help me! It is important to define what functionality is "basic". That is what will be built into the primitive network simulation. This will be the parent of all other simulations.

Okays, lets say we have a simulation of OSPF neighborship. Now what? We need to define states in the simulation which are possible real time scenario's. We need to build a model, a deterministic finite state machine.

Something has to be done dude.. we cant have millions of perl scripts running on billions of routers and zillions of software engineers supporting them!!! some sort of intelligence has to be built into the simulations.... but what and how..