Ashwin (ashwing@ieee.org)

topline and bottomline of Indian academic research

2011-08-11T00:11:00.001-07:00

In any business, there is a bottomline and a topline. I believe, that the bottomline of academics is to generate ideas that are of quality. Which means that the bottomline should be publishable, patentable quality research. The topline, is the impact to the society. I think these two, i.e. the topline and the bottomline are not isolated from each other. If anything, these should be tied to each other. Good research can be used to make a good societal impact. Not doing good research and yet having a societal impact, IMHO means that we are in principle reinventing the wheel. Which is also OK, but that this is not the work of faculty at an IIT, but say technology developers in CSIR, DRDO or other such labs/organizations. A faculty IMHO must develop ideas, and then if possible ensures that these ideas see the light of the day. The delivery apparatus for getting such ideas into the field and mass-adoption is clearly missing in the country and I would like to cover that. I think a three-way nexus between the government (funding body), the PSUs and the faculty is key to making such an impact. For example, if say BSNL/MTNL tells me that we need to build a large nation wide network. i then go to the government, and ask for funds, telling the govt that I will create the technology for such a network. The govt, pushes a PSU, say BEL, who then is responsible for productizing the technology that I develop. BSNL and BEL then are the PSUs. In such a situation, the technology I develop, is first tested academically, through quality journal papers. It is then prototyped in a lab setting, whereby it is tested by BSNL/MTNL to see if this is good enough for them. BEL then develops the products and is responsible for the large scale implementation. This kind of model was implemented by us with MTNL. I think it worked very well. The government funded the initial effort, MTNL funded their network, and IITB provided the technology. Revenue is being shared between IITB and MTNL. To me the intial satisfaction was the journal article that was published around the technology. The next level of satisfaction was the transfer of technology and implementation of the MTNL data-center using our technology, the third level was when users began using the data center for commercial use. The bottomline is clear to me - journal papers, the topline - the societal impact. From a faculty point of view, we can have people focusing only on the bottomline - classical scientists. We do hope to have people who focus on both the topline and the bottomline. But in my very humble opinion, we cannot have people focusing only on the top-line.

Our product and our relationship with MTNL has been covered in the news..

2011-07-10T19:27:00.001-07:00

http://www.thehindu.com/sci-tech/technology/article2216900.ece

video on our work

2011-05-04T20:02:00.001-07:00

http://biztech2.in.com/videos/networking/has-iitb-designed-the-worlds-fastest-router/107202

our work in the news

2011-05-04T20:01:00.000-07:00

Some recent news links:

http://www.moneycontrol.com/news/business/mtnl-launches-2-data-centres-to-open-3-more-this-year_539401.html

http://newstonight.net/content/mtnl-opens-2-data-centres-launch-3-more-2011

http://www.indianexpress.com/news/mtnl-to-roll-out-fastest-data-centre-developed-by-iitb/783626/

http://www.mumbaimirror.com/article/35/20110430201104281502453572f674a47/IITB-leads-the-way.html

time for a change....

2011-03-29T02:42:00.000-07:00

This is the right time for a change. A new change is anticipated, it will be fast, swift and efficient.

Cryptic, it might sound, but that is all that I can share.

launch of IndiaTelco 2011!

2011-03-03T19:16:00.000-08:00

The Gigabit Networking laboratory in its continuous efforts to facilitate broadband penetration in India would be launching a new 2-day workshop called IndiaTelco Summit. The scope of this workshop is to facilitate interaction between providers and vendors, integrators, and customers.

The website is at www.indiatelco.org.

product launch

2010-12-20T19:38:00.001-08:00

http://www.dnaindia.com/mumbai/report_iit-bombay-develops-fastest-data-centre_1483523

Obviously everything is secret about it.

More details would be revealed mid-way through the next fiscal.

Good job team and may your tribe increase.

An observatiion

2010-10-30T20:26:00.000-07:00

Those countries that have been at the forefront of 3G evolution, have not made that technology the main choice for broadband.

For example, consider countries like Finland (#1 broadband penetration), Norway, USA, or even countries closer to us like Japan, Korea or even developing nations like PRC. They are all certainly rolling out broadband 3G and 4G (on the horizon) services very soon. But, these are the same countries that have been most active in the Fiber to the Curb (FTTx/FTTc) initiatives. These countries are the ones that are making the most impact in terms of getting fiber to end users.

What does this say? To me it says that - yes 3G/4G is certainly a good option for the mobile broadband user, but one cannot do away with the fiber to the curb or even the end user, if one wants to have sustainable broadband initiatives.

This to me is a critical aspect of broadband policy for the future of the Indian scenario. We are tending to neglect the basics of telecom too much, opting for easily visible and short term impactful options.. leaving the more sustainable options in the dark..

The telecom road ahead..

2010-08-07T19:39:00.000-07:00

Telecommunications in india: The road ahead

Two events that have captured much of telecommunication news in the recent times are the 3G auctions and the suseptibility of Indian networks to foreign vendor equipment, especially those equipment with malicious intent. India represents perhaps the fastest growing telecom market, especially from the mobile perspective. A cause for severe concern which can lead to an information divide is the lack of high-speed broadband connectivity to end-users, especially in the residential broadband space. With less than 10% broadband penetration, the question one asks is what does India do to solve the broadband problem. This question requires further examination in light of the recent 3G auctions and vendor issues. In fact, the solution to the information divide and broadband penetration lies in the intelligent technology choices that would facilitate building up the telecommunication infrastructure and communication landscape in the country. To solve this problem, there is a serious need for reprioritization. We have to think long term and from a perspective that is in the interest of the country, business, citizens and the government. The need of the hour is to encourage practial and applicable research, innovation and production of battle-hardened telecommunication equipment that can stand the world onslaught from established vendors. Indiginisation of telecommunication equipment is perhaps the second most important aspect of indgenisation that the country must indulge in, just after defense, given the impact that information and communication networks and systems would have on the growth of the country. The question one asks is what are the technologies that we should be investing in? Should we follow and imbibe on products that are developed in other countries or should we begin an endeavor to support home-grown innovation. The answer lies in understanding the business opportunity of telecommunications in India. The growth potential of telecom and the business opportunity that this segment provides combined with the threat of assimilating foriegn gear have a strong implication on the way our country has to gear up for the telecom sector. The telecom industry has to play a thought leadership role in using ICT for the common man, in being able to streamline, scale and provide e-governance, e-trading and e-commerce to a larger section of society. Tough decisions need to be taken that empower the country into the next generation of communication methodologies. India represents a huge and unique market - the size implies a strong business case, while the uniqueness implies the need for indigenous products. Together, we have a huge opportunity to let our telecom industry grow. We have to examine what the government and TRAI can do to make this a success. Likewise, we have to cosnider the role of service providers and how they can be helped to move forward in roling services and empowering citizens with broadband while still meeting their bottomlines. Issues such as consolidation need to be addresseed in this discussion. To complete the telecom triad, along with the government and providers, we must also encourage vendors and vendor manifested technology, especially those technology choices that suit India the most. The discussion must focus on moving away from just a service offering based business, to a next generation technology business with a big dividend.

Perelman rejects award!!

2010-07-06T05:36:00.000-07:00

http://www.thehindu.com/news/article495428.ece

Great thing that Perelman teaches lesser mortals like us is about abstinence. From the context of sustainable peace, abstinence is absolutely necessary.

In academia for example perceived career growth happens when one does NOT abstain, in other words, one has to "apply" for promotions. To me this is demeaning one's righteousness.

Perelman goes even one step further in rejecting an award on grounds of ethics. Popular belief often propagated through the grapevine is that awards are often fudged. To get them on your own merit is a big deal, and to reject it further is even a bigger deal.

Hats off to this messenger of ethics and may his tribe increase!

rajneeti is a big hit in India!

2010-06-07T20:37:00.000-07:00

..but of course. A convolution of Mahabharat and the story of the GOP of India. It teaches us what we know best - to play games.

citations..

2010-03-27T06:55:00.000-07:00

Yesterday at OFC in San Diego, there was an informal discussion amongst our friends on citations. Academics believe citations are important - to me it all depends on who really cites your work. One of our friends - a well known academic commented that by visiting lesser developed nations and giving talks there, the citation count increases. Methods like this and others where only your friends or other academics cite your work are really crappy. I have noticed that most of the times academics in the area will cite your work. Most of these guys will do incremental stuff and will not be in any way expanding the horizons of your community. So what is the point in measuring the citations. To me, if someone from Cisco or NSN or AlcaLu cites my work, this is more important than 100 academics citing my work. Yes I am an academic too, but we are finally all working in applied sciences - where industry acceptance is the final goal!

3G auction in India

2010-01-30T07:42:00.001-08:00

Recently, 3G auctions have started about in India. After x1000 crores, or several billion USD investment by providers, my concern is whether providers would have any money remaining to roll out 3G services. From my rule of the thumb calculations, if a provider spends 750 million dollars in getting the licence, roughly, he would need another 2x or 1.5 billion in network wide investment to create the infrastructure. Even if the provider were to leverage the existing 2G or lower infrastructure and in addition just add 3G equipment only on a grow as you earn basis, it would still mean about a billion or so in investment for the equipment alone. The sad part is that most of this billion dollars of equipment investment would have to come in play within the first 1-2 years of deployment to make some business sense and lead to a market share of proportional value. Given the current situation that private players in India are in, it will hence be unlikely that we will see quality 3G services being rolled out in the short future. It is of course a different matter, that 3G itself will not be useful due to the high price points and the high population density.

What this means is that the ROI on 3G is likely to be minimal, and if anything, would be a deterrent for the introduction of new technologies.

In such a situation the money spent in procuring licences will go waste - the only way to recover some of it, would be to continue the phenomenal growth of voice services and roll those over in the 3G part of the spectrum as well!!

on the dual nature of life and technology

2009-11-16T09:52:00.000-08:00

Life seems like an extension of technology - in the sense that nothing is permanent in life and so is nothing in technology. You can do both well, if you do not get "attached" to either. You can make a terrible mess, if you are attached to either. And forbid if you are attached to both, then doomsday for yourself and technology is inevitable.

Now what does this mean for the telecom researcher.. wait and watch!!

Bharti MTN Deal

2009-09-30T22:01:00.000-07:00

What does this tell you: http://www.lightreading.com/document.asp?doc_id=182496&

To me it is evident that there is something more than what meets the eye. Did the SA govt deny the deal purely on non-technology reasons? I do not think so. To me, it seems that there was a strong underlying technology focus that went below this sentiment. Probably something that GOI missed out and SA found out. I know what that is, and I know it was good for telecom and our community.

Making news :-)

2009-08-23T06:06:00.000-07:00

Our paper on Broadband Services in India has been covered on the front page of "The Hindu".

The newspaper link is: http://www.hindu.com/

The actual article is: http://www.hindu.com/2009/08/23/stories/2009082355091000.htm

The original paper in IEEE Communications Magazine is here: http://www.ashwin.name/papers/commag2009-ashwin.pdf

Effect of age, time, workload and stress.. the worst is yet to come :-(

2009-08-01T04:59:00.000-07:00

Cracking the nuts and bolts of telecom research

2009-02-28T06:55:00.000-08:00

The commoditization of networks the world over is certainly something that we did not expect a decade ago. It appears now that there is far more manpower in this field available than what the traditional networking companies need. In such a situation the obvious question a young grad student would ask is whether one should invest their careers into this field. A short answer to this question is Yes - not because that is 'my' bread and butter, but because networking in general and telecom in particular have a lot to give than what we already have. I am going to try and pen my thoughts and pardon me if I bash up a few adjacent areas of telecom in the process. My goal is to bring to you, the young kids, the future of tomorrow - the expected virtues of telecommunications and this vibrant industry that we are all part of.

1. Think business: One of the things that always irks me is that the research that many of us do (including a small fraction of my own group), is quite disconnected from business. Ask yourself this question - every article/paper you write, every idea you develop, would a consumer buy it? More so, would there by a delivery vehicle that would be able to take this idea from you to the consumer? In short, would a provider/vendor/integrator harp on to your idea? An easy way to wriggle out of this question - is to simply say - we are researchers, we just want to do "research", commercial success is passe! Well, unfortunately for those friends who say this, the reality is that we are (whether we like it) into applied science - and the application is largely commercial. So again, if you are doing something that needs more than just a wheel to turn, then your eyes should battle multiple eyelids :). If you are one of those who says that hey I am developing technology for the future - assuming that X and Y bits of technology would be available, I am developing Z which would be able to meet A's "future" problems. You see the variables in X and Y, these are single points of failure. If you are going to invest all your efforts into developing Z assuming X and Y would happen, then ask yourself this question - what is the likelihood of X and Y not happening. More often than not you would not have the answer to this - simply because X and Y are innovations that would happen in different fields - across layers, or simply over time. Ask someone who has dabbled in X and Y or in earlier manifestations of X and Y about the likelihood of these happening. One of the things you would see is that people simply follow a bandwagon - those at the top of this wagon do benefit from their "blind" followers, but for a whole lot of people who are simply followers, there is a big problem - they cannot and will not succeed. This is when bubbles burst and a whole lot of our researchers are left jobless. Who is at fault? The wagon driver who started this or you who followed it? A simple sanity check can do wonders. I have seen many fields being developed that have had no impact on society, industry (as if they were two different things:)). This is simply a case of researchers scratching each other's back. The problem is that it seems we do not have the "guts" if I may use that word to simply stand up and say Hey - this is JUNK!

2. Have conviction: Many areas of networking that are these days in vogue are fictions of imaginations - their applications may be immense but their deployments are too scattered to be noticed. But in academia or in some cleachae industry labs (with no link to the business units) are being actively funded or pursued. These areas are generally trouble spots. Typically the people who drive such areas are veterans of "scratch my back" philosophies. A simple way in which a young grad can filter this is by corelating the work that happens in these labs or in universities with leading business dailies and supplements. If no major company has this work anywhere on their roadmaps, then there is something fishy about it. It is good to write papers in this area, but then, the moment you are done, you are gone! The papers are going to just be archived and refered by no one except your "scratch my back buddies" and if you are a wagon driver then the blind men (and women) in your wagon :). What I have observed, is that many of us, even in our grad days are able to understand that there is something fishy about a particular field - but unfortunately we do not have the conviction to take tough decisions. So, always have a roadmap of technology and business needs. For example, when I was done with my Ph.D data traffic was just taking over voice - some of us knew that video would be next and telepresence would follow. Of course we could be wrong with telepresence, but then at least video turned out to the a good stepping stone in our careers. It is important to identify bubbles - this makes you stay clear of rude shocks. Of course, if you are the one that created the bubble, all you have to do is recruit folks for your bubble - not a difficult thing since the picture can always be portrated to look merry. If you do notice that a large number of big names are big because of the bubble that they created. Ever wonder why nothing grows near a banyan tree ;).

3. Think big, think the bigger picture: At times it is easy to digress from reality. Often we forget the larger picture - we think that the work we do is the right path ahead. It is a depressing feeling to know that the work you did x number of years ago never left the papers you wrote. The question is then why did you do it? Obviously not everything you write can become products but an area in which you work certainly should be impacting business in some way or the other. When you do think about the bigger picture, also think about what are the blocks that remain between your work the the larger goal to society and to industry. My current head once told me - I am happy that you are working on large problems. It is better to work on larger problems even if you do not end up solving them completely than to do incremental work. I think he makes a lot of sense. The bigger you think the bigger your imagination gets - the better your ability to "scope" a solution. I am sure we all know our basics well and we can apply these to small problems as well as to larger than life problems - the latter need more than just the basics, this leads us to scope more interdisciplinary areas. That brings me to another important aspect of research - being interdisciplinary. In 2008 when I moved from the Stata center to the RLE labs at MIT, one of the things that caught my attention was a snippet across an elevator - it says that the last century was about innovation, this one is about collaboration. The Internet of course has been an excellent medium for all sorts of collaborations! But ever wonder what are the compelling reasons for collaboration? It is interdisciplinary nature of the business. When I design a telco box, I jolly well need to understand the mechanics, the algorithms, the optics, the RF, the thermodynamics and even the acoustics! A wide variety of expertise, which is hard for one man to gather in a short lifespan (rather say a short shelf-life - more on which follows). So we need to be interdisciplinary and hence collaborative. Again, if you do not think the big picture, you do not need to be interdisciplinary, you do not need collaboration (unless all of your collaborators are the scratch my back kind ;)).

4. Enhance your shelf life: One observes that researchers come and go! Some stay for a long time and are active. I often wonder what happens to some really smart people - who perform very well over a short period and then fade away into oblivion. What happens to these really smart guys? Do they get burnt out? Do they loose interest? Or are they pushed away? It does seem to me and I could be wrong here (as I have not yet proved that I have a long shelf life), that many of these folks are victims of their own legacy: (1) solving non-important problems (2) being adventurous in a single area of expertise (not being interdisciplinary) and (3) Not being on the right wave at the right time. The first one is something that affects a lot of folks whose shelf life is short. They work on problems that are disconnected from the real world. So while they do solve some really interesting (non-real) problems, these problems usually end up in huge energy drainage. The thing is that many non real problems are stemmed out from funding agencies which want to "investigate" into multiple areas hoping to get the right tact in shape. Don't blame them, they want every aspect of the science to be well covered. The second problem of interdisciplinary work is legendary. It is tough to be interdisciplinary. But you have to take that chance. Finally, heard the phrse - fortune favors the brave - well thats all about being on the right wave at the right time.

4. There is always another day: One of the keys to solving the shelf-life problem is that there is always another day. Research unlike life does give you a second chance. It can be quite forgiving. If you take that as the bottom line - I am sure you are going to do well. One should not worry about taking risks. Risks are like bursts of learning. Each risk gives you an ability to learn, and you can grow with these. In an increasingly competitive world, it is often best if you took your risks with conviction leaving competition to the wind. Each failure will open a new leaf and if you are true to your research, there is always a way out - there is always another day.

5. Be known by something: Often there are many interesting problems that we can solve or want to solve. What is important is that at the end of the day, month, year, decade, you be known as someone by something. What this means is that you should be known for your work on something. So while it is great to be deep and all, it is many a times another thing that you do work on a larger problem - a problem that you solve for which you are known as. So one has to ask the question - what is Mr. X known as? Someone who knows the intricacies of the field Y or someone who pioneered Z that was useful in many way. I often see great names - people who have a large # of papers or who are supposed to be "stars". Many of these are stars because they belong to a certain pedigree, many are stars because they have a large number of pubs, many are stars because they have led the community in many conferneces or are editors of multiple journals, but only a few are stars because they have pioneered innovations in the community. To me, this is important. I would personally be want to known by something I have done technically, rather than my pedigree, number of papers, or organizational abilities.

Many of these thoughts are not really mine - they have come to me as a result of interaction with some of my older colleagues - they have seen it all ;)

Enjoy.

Omnipresent Ethernet Test-bed is up!!!

2009-02-26T18:18:00.001-08:00

Our latest test-bed is up.

Its called Omnipresent Ethernet. An article will soon be released as well. Kudos to Saurabh and Raviraj for making this happen. More work to follow..

Broadband in India: Sparse and Farce

2009-02-26T18:00:00.001-08:00

We have just released a new article on Broadband Services in India.

http://www.ashwin.name/papers/bb-india.pdf

Enjoy :-)

Ashwin

Compassion..

2009-01-30T19:40:00.000-08:00

Was reading a book, "The Buddhist Path to Simplicity" by Christina Feldman, and came across this interesting passage, which I quote:

"Compassion springs from a mind and heart deeply rooted in simplicity, integrity and a profound understanding of the interconnected nature of all life. Compassion is a transforming quality of heart we cultivate, nurture and refine. It is rediscovered through the falling away of the layers of fear, resistance and anxiety that have the power to veil the innately compassionate heart. Our challenge may not be so much one of becoming more compassionate but one of learning to let go of the clouds of confusion that obscure the powerful compassion within us."

Spend a moment, thinking about all that we are agitated about - and one realizes that when we let go of these things that we crave for, we are often better off.

2008 - chequered field..

2009-01-06T18:22:00.000-08:00

2008 ends, and this is a time of the new year when things are slow, sluggish and even boring...

Research in 2008 was very good, some good ideas, some nice new results. Our group performed reasonably well, close to two dozen articles, which was quite exciting. To add to that, we had 8 US patents issued.

Whenever there is a rececession, there seems to be more vigor in our research, I guess the insecurity sets in and we perform that much better. This recession though may be different from all previous ones that we have seen. THe mantra seems to be focussing on operational expenditure based models - no capex for providers and this seems to make everyone except the customer happy!

A goal for us in 2009 would be to try to develop models that outline the problems with the zero-capex model. Easier said than done. The models are hard to come by - Nash's equilibrium exists both ways and that seems a tough sell. But we will try.

Another area of improvement is in the network architecture domain. The next generation box in the metro is likely to be able to understand the application - whether it is cloud computing or data centers, the boxes will largely influence the design features. How can an application be tied to the data and transport layers, and further how does this arrangement scale. Once again, Ethernet is the key to success and good commercial deployment.

2009 promises to be a year of uncertainty and challenges - research wise this means a year of opportunity, and I hope our group makes the most of it.

Las vegas data centers :-)

2008-10-15T15:39:00.001-07:00

Writing this from a panel in Las Vegas - the weather is beautiful and the land here is cheap - ideal for cloud data centers.

A recent article on the comments I made are here: http://in.news.yahoo.com/48/20080928/804/tnl-iit-prof-urges-expansion-of-optical.html

and here:

http://www.expressindia.com/latest-news/iit-prof-urges-expansion-of-optical-fibre-network/366776/

Look at the similarity in the language and gauge what you can!

PBB TE and T-MPLS

2008-10-11T09:56:00.000-07:00

Some serious discussion going on out there comparing PBB TE to T MPLS and who will rule the roost of metro Ethernet. While our group has been big fans of PBB due to its simplicity to play around with the VLAN tags, it does seem that the other camp has some good things to show off. Among others, the PW part of T-MPLS using Ethernet PWs may lead to certain layer 3 advantages that are as of now seeming impossible using any other technology. That gave us some shivers in the past few days..

Moving back to he financial crisis - it seems that is one more reason why the cloud would happen - there is no capital expenditure (on one hand), and to come out of the crisis we need to process more information, create better applications etc. So can we sell the mantra - build the cloud to enable a financial monsoon?

Some nice cloud names: NIMBUS: Network Integrated Metro Bus for Universal Services
STRATUS: Service Transparant Reconfigurable Adaptable (Available) Transport for Universal Systems
CIRRUS: Cost Intensive Recongurable Reliable Universal Services

back to basics..

2008-10-07T22:30:00.000-07:00

Spent 15 days without any unwanted travel - getting back to the basics of research.. feels good.

On the same line of thought - strip all of Ethernet's powers, kill STP, remove IPG, simplify the PCS, make the MII more user oriented, simplify the SERDES in the PCS - things should become easier.

Bob Metcalfe spoke at this year's OFC on terabit Ether. While that seems a way off, one of my students investigating 100 GE, seems to be convinced that the 500 odd parallel lines that run in from the MAC to do 100 GE is an overkill. And I do agree with him. It seems we have come along in a circle - started with packet switching, WDM, GigE, cloud computing and now to sustain this cloud we need a better way of doing Ethernet.

On the same line of thought - we keep reading and hearing about "proper layering". To me that is real BS. To me that means that people who do not know about other layers can safely survive in their cozy jobs by doing proper layering. In fact layering is an overkill - time to redesign the damn network..

Customization issues

2008-09-19T17:39:00.000-07:00

IT managers are likely to think that the cloud is not permeable in terms of launching operations for reasons beyond security - customization. Can two companies who's IT mantras are contradictory use the same cloud, and if so how would that affect the functioning of the companies as well as the cloud.

The short answer to this is that, yes, certainly companies with diverse IT philosophies, and hence applications can, and will use the same cloud. The good thing about the cloud is its ability to support fundamentals of distributed computing - communication and computation at a very fine granularity. What this means, is that put anything on top of the cloud, and it should give you a silver lining!

Of course this also means that we need a strong middleware that would be able to abstract, deposit, retract and process application granularity into processing and communication granularity. While the computation or processing part is well researched and worked on - through the ages of parallel processing, the communication part is what interests us, and we have some pretty nice ideas that make that happen.

The communication part aka virtualization is well established through our OE-model (optical+Ethernet), and it should be able to meet the requirements of generic cloud computing.

My gut is that we do need the control plane to be common - between the computing and the communication layers. The control plane has to have horizontal integration (ability to virtualize) and also allow vertical integration (ability to process distributed application chunks) in a pragmatic and efficient manner.

latency issues in cloud computing

2008-09-17T05:11:00.001-07:00

Someone was commenting about latency as a potent issue that could hamper cloud computing. My argument is that, what are you comparing this against? If you are comparing against an ERP app and you are remote, then the latency you would get across a cloud would be similar to the latency you would get across to your organization - both depending on the access/metro and core network that connects you to your hosts.

This is a good enough reason to then go for cloud computing.

A personal memoir'

2008-09-15T14:16:00.000-07:00

To him, who I never saw, but without whom, I would not have been... my late father - Dr. Anil Gumaste

Today being his death anniversary - thinking of him, and saluting a well accomplished neurosurgeon.

DCE and latency issues in the cloud

2008-09-13T06:00:00.000-07:00

In my last post I spoke about the importance of Ethernet. Having withstood every layer 2 (and even layer 3) onslaught and coming out with flying colors, Ethernet is certainly the way to go but for DCNs and clouds there is a lot of optimizatiion that needs to be done.
Now think about large corporations whose life line is the IT operations that help enable convergence as well as integrate. If you want to put everything on Ethernet then the key to that is to make it really carrier class from end-to-end. There is a lot of work done on making Ethernet carrier class - but these solutions are only in the metro area or may be if you extend those, in the long haul as well. But when you talk of enterprises, the carrier class features should be within the campuses as well. wonder why there is no talk about carrier Ether in the access - a space that would have a lot to contribute to in terms of application and hence revenue as well.

Data centers, Ethernet and optical transport - key to cloud computing

2008-09-06T06:20:00.000-07:00

The way things appear, from a networking stand-point the 3 key technologies that make or break cloud computing are data centers, Ethernet (in various manifestations) and optical transport.

Data centers are the brain that one can say is being outsourced! In order to facilitate availability of processing power across the web, we need network attached intelligence with added computational, storage and collaborative abilities - all of which are well juxtaposed by data centers. DCs can be of 2 types, the stand alone one or the distributed one, and given a good communicative infrastructure, either of the two should not matter. The intra DC design itself is a subject of tremendous introspection and research and a lot of work needs to be done here as well. Especially if one is to build these using server or even desktop farms. Which means that there are issues at higher layers like file sharing and file systems itself that need to be optimized. Yahoo's hadoop is a good mechanism to do the former and is well recommended by a large number of non-profit organizations.

The second key technology is in the transport of data between the DCs, the DC and the end user and between multiple businesses (DCs again). While not much attention is being given to this - relegating this area to the likes of equipment vendors, it should be sufficiently be clear that this area is an area of strong growth as the networking requirements are far different from anything we have seen before. Voluminous, latency sensitive, dynamic bandwidth needs cannot be met through dedicated circuits or unpredictable packets. We need something better, something with a packet nature but a circuit type reliability. Ethernet in the metro and long haul is likely to be such a candidate technology. Making Ethernet dynamic, reliable and available will be the key for facilitating efficient cloud computing across large networks. It is quite unlikely that the present variants of carrier Ethernet alone could scale, in fact, they might be best suited to be frameworks to be able to facilitate something more powerful - a new Ethernet pedigree. In our CAMPUS solution that we are presently building one such approach is being considered. But this may not be the only way. More on the CAMPUS solution in a few weeks from now.

Finally, the underlying transport technology is important for the cloud. From a stack perspective the physical layer is simply something that would carry bits. At least that is what most CS academics would tell ya. But the real thing is different - the phy layer can be an important weapon in the DCN arsenal. It is the single reflecting point in the network - allowing functionality to be cross communicated. The phy layer having the primary function of transporting huge chunks of data from one point to another, with minimal errors is naturally composed of optical fibers. But if we think about the added requirement on the phy layer, we realize that these optical circuits need to be reconfigured dynamically in two ways: (1) the optical circuits themseleves are to be set up and torn down as part of a fast moving virtual topology. (2) the circuits may have different line rates which change rapidly. Moreover these circuits are subject to particular data center requirements. The optical layer hence can play a key role in meeting these. Some of the technologies we have developed are well suited to this - see my CAVALIER paper.

I read this interesting blog on Ciscos website on DCNs and the way things are moving.

I was also following up on Nuova's acquisition (by Cisco) and what that means for the data center business.

People are getting this right but there is still a bit more to be made, and being bottom heavy may be the way to go.

service as a business model

2008-09-03T18:52:00.000-07:00

A significant amount of growth over the past one and half decades is based on the services sector, both in developed and developing countries. The services sector now is trying to break into traditional boundaries which have been well guarded by hawks of the products based business model. Cloud computing is one such example where software products would be replaced with software services, and one would not need to purchase any more expensive software. All one would need is a graphic rich front end device with a good bandwidth pipe. Of course getting a good bandwidth pipe in the US is fairly simple and inexpensive, but doing the same in India is next to impossible under the current bandwidth plans. Getting back to the services sector, there seems to be a percolation of this model from software services to non-traditional domains. Amongst these, the ones that affect our business the most are telecommunication services and computational services. Cloud computing of course is the merger of these two, brought to the fore as a singular entity that enables the end-user to be able harness the power of on-demand computing over a wide area network - the Internet. Amongst others, an important benefit of this seems to be in the Enernet sector - saving Energy - the thoughts on which i will soon write about.

The metamorphosis of the telecom industry from a product oriented to a service oriented one is both a good and a bad thing for business as well as technology. In the short span of time, providers would no longer buy equipment but would simply give a fraction of their earnings to the vendors in return for deployed "capacity". Vendors would hence no longer have to sell equipment simply using equipment as services arm that have been offshored to the provder. This model which with audaciously can be described as a "zero capex" model has the potential to revolutionize the telecom industry. Naturally, the question one would ask is whether this is a good thing for all of us. From a numbers perspective, it gives the provider a risk free network - no investment on equipment, no investment even on maintenance. Just share your revenues with the vendor and the management company. To the vendor this means lock-in. Once you are in, you are in for a while (into a provider's network). The question of course is whether this model is good for the consumer as well? i frankly have my doubts about this, and which i am working on trying to substantiate through theory and simulations.

While this is one example where traditional business has become one of a oriented nature, there are several other examples popping up all the time that transform technology through different business models. Again, from a short term perspective, the quarterly reveneues look good, the balance sheets do show black ink all over, and there is far less risk than one likes to work with. But the question remains whether these models would work on a longer term perspective. The questions for which i am looking for answers are the following:

Does the consumer benefit?
If the consumer benefits then does he benefit always?
What about technology growth? Is that taken care of?
What about innovation?
Can a start up ever be incubated successfully in such a situation?
Will the government have any regulatory role to play?
What kind of regulations are needed to make the services sector foray into traditional product economies?
Does this model scale?

I hope in the next dozen weeks or so, I am able to get answers to these questions.

Rehashed cloud computing

2008-09-03T06:06:00.000-07:00

Can one distinguish clearly between cloud computing, utility computing, application oriented networks, grid computing? To me these are variants of data centers (either uni-located or distributed) accessible across a network. The fundamental question then is to understand the requirements from a networking perspective on these?

Needless to say, that the underlying network must support one or more of the following requirements - ability to move large chunks of data, to move the data in real-time, to enable virtualization of resources, to be able to multicast instantly and finally to be able to be secure. Whatever be the front end of this - the back end must be a combination of optical and Ethernet technologies. Very little work is done to make the back end support pure cloud computing and my gut is that if we merge communication with computing the resulting platform would be a very smart entity that would help collaborative innovations happen.

A larger problem is to make this virtualization happen on mobile devices, end-user platforms and for consumer driven technologies.

We are now investigating what we can do to make cloud computing happen. Both from a HW and SW perspective. I don't think the larger guys have got this right - the business model is still not there. I read this excellent article by Aaron Weiss called Computing in the clouds. That tells you exactly where the business model fails. So while the amazons and googles of this world are doing things, it remains to be seen how much they would succeed. I foresee that network vendors especially those who have diverse portfolios may have a good chance in biting a large chunk of the cloud computing pie.

There certainly is money to be made, but how to do it is the question. also, it is quite likely that cloud computing is one area of technology that would be plagued by policy issues (especially those related to content security).

a first entry

2008-09-03T06:03:00.000-07:00

While, blogging may have been around for a while, what intrigues me is the business case for blogs, and the dissemination of information that can potentially lead to collaborative development. The theme of this blog is to enable collaborative development and collaborative innovation.

Ashwin