a mARTIAN dIARY

A while back I had written a post Flowers Vs Servers pointing out the similarities between the demand for flowers and the load on servers catering to educational institutions. I had ended the post concluding that since the former was a comparatively old ‘solved’ problem, we could try and extend its solutions to the latter. This required for us to understand how the old problem was solved and this in turn necessitated talking to the flower vendors and distributors.

For months this laid low on my list of things to do (along with a LOT of other stuff) but by providence rather than intent, I was able to do exactly that last week. While waiting for bus at KR market early morning, I glanced across to the countless flower vendors on the street across and suddenly the to-do list in my mind got re-ordered.

I immediately crossed the road and walked up towards a person who looked to be in his early thirties and a bit free from the selling (and the ungrateful wretch that I am, I forgot his name….thank you Mr Anonymous for all the help). To cut a long story short , both our verbal and non-verbal communication skills were tested to the extreme, so much so that, I am confident of cracking any dumb-charades question that even the most wicked of minds can cook up.

Anyways, among the stuff he said (or the stuff that I understood him to be saying) one idea particularly struck me as applicable in our Server case. Recount that our original crunch times came during festivals. Now think of a regional festival like Onam in Kerala, or Uagadi in Karnataka and Andhra or Ganesha Festival in Maharashtra (this is arguably not a good example). Now in all these cases the demand for flowers is very high, as flowers are one of the most important components in the celebration. BUT the demand is regional, so a solution to meet the excessive demands would be to import from other regions.

Now translating this to our scenario, let’s think of the results of different exams of different colleges. They are usually announced at different times, so in the analogy terms, the ‘demand’ for bandwidth is ‘regional’. So a solution similar to the ‘import’ solution would be to link up the servers of the various institutes using a common backend and share their resources for showing the results. By doing this they will be able to cater to the large traffic without increase in their infrastructure.

Of the solutions I heard this was the only one I was able to translate. Others included selective farming and also use of refrigerators (cold storage) at the retail level.

Recently during an internal class on Innovation, I was inspired by a story of how self cleaning windows/cloths for sky scrapers were inspired by our very own national plant The Lotus. A classic example of “learning from unusual sources”! It was an instant hit with my mind, which loves to draw analogies.

Fast forward to present, as I was driving to office in the morning, I could see ample signs of the grand Durga Puja festivities that were concluded yesterday. During my “studying” days  Durga Puja or Puja Holidays has always be something that I looked forward to since it meant a 10 day holiday for us KV’ties and also no studying for 2 days where you would do “aitha puja”. After moving to Bangalore, the word “festivities” have had another scene attached to them in the sequence of images that one attaches to a word, a image of a myriad of flowers, cacophonous bursting of crackers and the vivid lighting associated with them. Not to suggest that we don’t have that back in Kerala, but here it is stretched to the limits (Yes I have not watched thrissur pooram which my friend assures me is much bigger) , probably due to the closer proximity of flower growing regions as well as fireworks producing region (the latter I am not sure about).

Anyway, as I was navigating my way thru a thick colorful patch in the road, which till yesterday was embellishing someone’s beloved vehicle in a different form factor more akin to a floral pattern than the thick muddy paste that it was now, something stuck me.

The demand and supply relation of flowers!

Imagine the normal demand for flowers is some value of X Kg per day. This ‘normal’ demand would be constituted largely by demands from temples, homes for decoration (of puja rooms and framed photographs alike) and lastly but most importantly (from the sustenance of South-Indian race point of view) aphrodisiacal decoration of tresses. Now the next local maxima in the demand vs. supply plot would come on Sundays and other auspicious “marriage” days, which may push by the X by a factor of 3-4.  And the absolute maximums would come on festival days where the demand can spiral up to 10-20 X.

Now, take another look at the commodity here, its flowers, a sweet smelling, high on water content, easily perishable commodity, which has a shelf life of 1-2 days. Naturally this question of meeting demand was not asked just yesterday and the growing patterns have emerged and other supply chain techniques have evolved to solve it. An inquiry would probably give us more insight to the patterns and “innovations” that this has brought about and if one looks closely enough we might be able to abstract and get the patterns for the same.

Now coming back to our familiar world of 1’s and 0’s. Consider a server that hosts a college website. On a normal day the traffic would be X. On the day that an assignment is to be submitted (assuming that it is given online) the traffic would increase to 2X-3X with the frantic downloading of the pdf’s before the deadline. Again on the day of the result announcement (say of a test like CAT for MBA which 1-2 lakh ppl write each year) the traffic would be (number overflow) * X.
 
Same would be case of a site streaming online media (cricket for example) on the day there is an Indo-Pak game and all the poor engineers are stuck at office (assuming that the web administrator is kind enough not to block the site)

After seeing the similarities in the two situations, can’t the solutions of the former be explored to get a solution for the latter?

Recently my friends (Vishwa and Vijay)  and I ( I know “I” seems odd but MS Word wont let me have it otherwise and yeah I know its PROPER English :P)  were discussing a problem that that’s having a detrimental effect on the sleep patterns of a lot of engineers and managers in the mobile phone industry.

The Problem can be stated as “The Missed call facility along with the Caller line identification feature is being misused by the customers by using the same to avoid making revenue generating calling.”
The main challenges accompanied with the problem is that

1. The problem in itself is not explicitly illegal/ unethical from the customer view point to solve by any legislation.
2. Any penalty/cost to be imposed on the missed will also be inadvertly penalizing the people making “genuine” missed calls.
3. A lot of customers with intent to make revenue generating services are not getting access to the system which in turn is costing the company invisible revenue loss.
4. The solution for the 3rd problem requires investment for a larger infrastructure than what is not financially justified by the actual “active” customers in the network based on the current revenue model

Now to apply the TRIZ contradiction matrix to this problem, it needs to be stated as a contradiction.
There may be multiple ways of stating the problem as a contradiction but I take up this specific way -:

Say you improve reliability of the network by limiting the total number of users as stated in the 4th challenge then the worsening quality is that the quantity of connections that can be offered with the same infrastructure is reduced and consequently revenue is affected.

Or in TRIZ terms it can be stated as

Improving Feature: Reliability Vs Worsening Feature: Quantity of the substance

This is a very abstract way of representing the contradiction but TRIZ is after all an abstraction
The solutions can be looked up at  http://www.triz40.com/
Solutions are
21. Skipping

• Conduct a process, or certain stages (e.g. destructible, harmful or hazardous operations) at high speed.

28 Mechanics substitution

• Replace a mechanical means with a sensory (optical, acoustic, taste or smell) means.
• Use electric, magnetic and electromagnetic fields to interact with the object.
• Change from static to movable fields, from unstructured fields to those having structure.
• Use fields in conjunction with field-activated (e.g. ferromagnetic) particles.

40. Composite materials

• Change from uniform to composite (multiple) materials.

3. Local quality

• Change an object’s structure from uniform to non-uniform, change an external environment (or external influence) from uniform to non-uniform.
• Make each part of an object function in conditions most suitable for its operation.
• Make each part of an object fulfill a different and useful function.

Now the theme among most of them is the break up the service especially looking at the  rule “Change from uniform to composite (multiple) materials.” (The quantity here being the number of connections as defined by us earlier) or from all 3 rules in “Local Quality”.

Now translating this into terms specific to our problem we can come with with- :
Provide different services with different QoS but at corresponding costs. As in, provide two or more separate networks, one that has a poor QoS but also has a much lesser “infrastructural cost” to be levied from the customers and provide another high quality network where the cost of service is higher but QoS is better.

This may at first glance seem to penalize the people who would be responsible for generating revenue even in the first case. But only by going into the actual financial details can we ascertain whether these new networks would exist at a higher cost segment or in the similar band as to now. (The lower QoS network might even get cheaper)

In one way I had always this solution in mind and probably that why I was able to see thru the esoteric solutions given by TRIZ, but then again innovations ultimately has to come from the Innovator: P

 Now the funny part -:  This already exists! In a way that is…

Take the case of Bangalore

There are three prominent GSM providers -:
Spice Hutch and Airtel.

For people who have experience with the networks they can see the semblance between spice with the first network (extremely low cost but not that great QoS), Hutch ( Low Cost and decent enough QoS) and Airtel ( the costliest of the lot with probably the best QoS)

One of the biggest takeaways you can have from the number “42” is that its not always the answer that difficult.(even though it did take millions and millions years of a computing time of the (edit) 2nd most advanced computer to answer it…still) . But after being exposed to TRIZ and Edward De Bono and his crazy hats, what Douglas Adams said makes a lot “more” sense.

When one works in a company that’s situated in the “other” part of the IT heaven that is Bangalore, it’s much harder for one to understand certain problems that run in the blood of every other bangalorean especially the much maligned traffic problem. Till I started making my now-routine weekly trips to Koramangala, traffic to me was like the demon that existed in my house but in the “that” unused room which I never had to visit. And in some crazy way it more like a kind economic stimulant to keep all those RJ’s and others in the radio industry employed than an issue.

But after coming face to face with the demon I have had to change my opinions. And with Murphy’s help, on the day I am late, it would take me 40 min to reach my destination and on others when I try to factor in the 40 min, I reach 20 min early with nothing to do except swat flies for the time I “saved”. Soon enough “traffic” slowly took a prominent place in my list of lunch break and ice break(er) topics.

But then one day it struck me. What is the most annoying part about traffic?
Is it the amount of time it takes or extra time it takes? Hmm…not really cuz if I were properly equipped (cue an ipod or a sony walkman) it doesn’t matter much.
Is it the time you waste just getting out of a signal to be caught again at the next one?
Yes this is annoying, but it’s probably not the most annoying or in a larger senses just a side effect of the main problem.
Then it dawned upon me that it’s the unpredictability that most annoying about the whole problem. It’s the fact that one day it takes 40 min and one day 20 min for the same distance, even without a major disruption like a accident or a fire(?). Plainly cuz the random traffic was more that day or I was unlucky to get caught in all the signals one by one. This brought about a paradigm shift in the way I viewed the traffic. It was no longer about “How to reduce the traffic?” but more about “How can I make traffic delays predictable?” and soon other solutions that I had not thought about till now or did not give too much utility points started emerging.  The important thing is for the frist question its hard to find a scalable solution where as the second problem has very scalable solutions.

I do not say that I have found a permanent solution or albeit even a temporary since most my ideas where invalidated by my friend during our discussion. Also I understand that what annoyed me, the un-predictability, might not even be of minute concern to others, but I feel that by redefining the question I was able inject a fresh life to the hackneyed approach existing to the problem.

The origins of usage of computer for typesetting of documents can be traced to the legacy of the veteran iron war horses known as typewriters. While computers were being developed for number crunching scientific (and military) applications, cranks with an eye for money minting realized the potential of this semiconductor baby to revolutionize the publishing industry. Thus was born the first word processing software and its connected professionals, born-again typists. The potential of this young prodigy was only superseded by its ruthlessness in outperforming and thus eliminating its competitors, which can be testified by the relic status given to typewriters now. Except for the aged writer, who is too stubborn or superstitious change his trusted tool, we will be hard pressed to find a typewriter anywhere else. But just as in the melancholy of defeat comes new learning; in the heavy march of victory valuable lessons are trampled without witnesses The tremendous success of word processing software over its predecessor lead to a ‘no prisoners’ strategy, where in most of the focus related to word processors was into excelling its predecessor. That is focus was totally on creating the ultimate tool that could handle the whole processor or word-processing ‘for printing’ child’s play. The italicized words convey the focus! So even when computers started becoming so common and cheap that, it was easier to read these documents from the system rather than a print out, the old tools were reused. So much to prove the laziness of man! There was motion in some direction leading to development of PDF and eBook formats. But still they lingered around the age old principles, too afraid to go for a radical change. It is needless to emphasize that a hardcopy of any document has its own advantage, at the countless ways it can be physically positioned, the psychological inertia coming from reading of books during the academic years and easy navigation up and down at the twist of a paper. But along with those lie its various disadvantages too. It presents information is a mundane non interactive way. The linking between the different concepts is usually inferred by language constructs or formatting. The occasional picture in the document does help reduce this disadvantage to a level, but a part remains. On the contrary, reading from the system monitor offers no flexibility as far physical positioning is concerned, except for some new age paper-like display devices but they can be a case or taking the river to the horse rather than vice versa. Also it does not offer (at lease current reader GUIs) much flexibility in terms of navigation that a book affords except for the search feature perhaps. And where paper is weak, the system gains a strong hand, its excellent environment for linked interactive interface for data from a book. It can present info the exact way it’s mapped in our brain. This brings us to the concept of concept mapping and related software. Watch this space