Being in the networking world, the only thing that I have been hearing about over the past several years has been the term "packets-in and packets-out", day in and day out. When I get to work, invariably the hallway tech talk will be about why packets are not coming out or why packets come out at a rate less than predicted or why the packets coming out mismatch with the packets coming in and so on.
If you are not onto internet technologies or routers, the only two things that you need to know is that any networking device (router, switch, etc.,) looks up for a destination address and sends it across accordingly. In other words, a lookup is made for the destination address and the packet gets routed until the final destination is reached (if you use windows machine, go to command prompt and type "tracert www.yahoo.com" and you will see the routers through which you get to yahoo's webpage. You can obviously try it for different websites as well). It kept me wondering. I think, we as humans, fundamentally do just that. Ofcourse, we bring in a lot of associativity to the contexts. But, at the lowest level, all we do is just store and lookup information. One could contend that we do perform processing of information as well, but when I think about it more, I think processing can in turn be categorized into a bunch of lookups based on information already stored in our database. In networking world, the one thing that most people work on is, how fast we can do lookups using the least amount of storage and thus cost. Just a rationale to the human context is that, our brain stores all the information that we care or we dont and processes items so fast using the least energy (when compared to a machine). Just a reality check on how far technology is off.
Wednesday, December 06, 2006
Looking up at Lookups
Posted by Suresh Sankaralingam at 10:19 AM
Subscribe to:
Post Comments (Atom)
Posts
8 comments:
OK..routers,packets....
searching my lookup....
routers...thiruvaluvar route bus...
packets...savoury packets...
brain...lookup....
what? brain...what?
Like the brain forgets unwanted things (according to the human importance), if the network can delete message which are seldom used and make space for latest ones.. then it can be fast (with least amount of storage).
Now who defines unwanted things to network/storage ?
Routers/Machines, if designed correctly, can do one functionality correctly and with uncomparable (wrt to human brain) speed.
We are still trying to model some of the things brain do and it is still too far away.
But eventually all the things brain can do with some process will/can be done by machines in much better way.
Real difficulty is in making the decision whether to type a response to this blog or work. Machince, if programmed, would've used priority and we do based on our mood :)
I think it is all a question of how performance is rated... Is performance measured by throughput, latency, power, area, or all of the above?
One of the analogies that I can think of is that, machines are like offshored jobs. It does lesser thinking, does same thing again and again efficiently at a cheaper cost, probably occupies more area as well.. Is it brain's solution to its own productivity to do more innovative stuff ? I do agree that it is a matter of time before brain can unravel secrets of its own...
hmm interesting...okie here is a fundamental confusion that I have...thinking of routers and machines, basically they perform the same function...1000 routers will do the same thing as 10,000 routers (albeit with some fancy stuff added)..however consider this statement (am not claiming it to be a fact)..."every human being is different implies that every brain is different"...can we really make this statement??? coz can one not argue that the human brain is basically processing information given by the user (the owner)...the same applies to a router or a machine...it does what it is told...now if this is the line of attack then I guess a strict comparison is possible...on the other hand if one is to argue that the brain is heterogenous and it depends on how a particular individual wants it to behave, then we cannot make a comparison..
hmm..okie this is kind of confusing but let me try to explain...as mindframes pointed out there is a significant amount of chains running in the human mind when decisions are being made but a rather toned down packet switching alogirthm works as a constrained optimization problem...now for switching algorithms we can always (in theory) achieve a global optimum and it is standard (say in reference to time consumed) for all routers..however for the human brain what might be a global optimum for one person need not necessarily be the same for the other..
therefore i guess the lookup function is like a pre processing task which is rather common to the brain but the value addition is in the actual processing...imagine a world where every human being thinks alikee...omigosh that would be so boring...now does this make any sense at all????
My view is that, any processing can be defined by a series of lookups onto the brain-memory database. And, brains differ in the chain of lookup-sequence that it does though most of them may contain all the necessary information...
@ mindframes: okie from this perspecitve what i was saying is as follows...
Stage 1: the brain looks up the information from the database. this is the common element. For the moment we can assume that this is homogenous and learning is not a problem. this is strictly for comparison purposes with routers because learning can happen even with machines (u can program in all kinds of new thingys right????).
Stage 2: the processing on the information so obtained is heteregenous. This is where the difference is. Rather this is what makes us superior over machines.
what say??
Very much agreed...
Post a Comment