RIM >> "Research In Stasis"

From Good Morning Silicon Valley blog entry Betty Ford Center braces for wave of CrackBerry withdrawal cases,

Millions of CrackBerry addicts may soon find themselves disconnected and yearning for a fix, now that a federal judge has denied Research in Motion's request to halt the proceedings in its legal spat with patent firm NTP ... "Research In Motion considers changing name to Research In Stasis".

I guess no Canadian technology success can catch a break ... They turn into [fill any word you like] with time. Will RIM be next Nortel?

Riya may show pneumonia in x-ray

I wrote in my last blog entry Riya in Healthcare about Riya's automatic photo recognition and its potential in healthcare.

I came across the blog entry Automatic photo recognition: Tell me if this chest x-ray shows pneumonia at a blog maintained by Dr. Vesselin Dimov. I guess I am not alone in thinking about the attractiveness of Riya's technology in the healthcare market. It is great to read some validation from a healthcare professional for one of the potential applications of automatic photo recognition. I see at least four additional applications of Riya's technology in this space.

Unfortunately, every body is focused on RSNA 2005, most probably the largest healthcare event, next week so I don't expect any feedback for another couple of weeks.

Riya in Healthcare

Recently, I read about a photo search startup Riya.com (See, The Birth of Riya.com at Peter Rip's Blog). It is an interesting startup and more interesting is the marketing strategy they adopted - a good case study how web logs can be leveraged in promoting a company, product or service. And I guess by writing this entry, I became another cog in their wheel.

My interest in learning more about development with Riya's technology is little different. Usually, when, I read about new products and services being launched focusing on web, I try to explore how they can be extended in to enterprise or incorporated in to vertical industry applications.

Reading about Riya actually crystallized a thought, I had during flickr acquisition by Yahoo!. I have been reading several blogs and comments about how Riya can be used for face recognition/comparison and for security purposes.

But I see some potential for such technology in healthcare industry. Over the weekend, I refined these thoughts and the specific problems it may address. I was going to write a blog entry describing these ideas. But as a fact-checking before writing, I decided to take these ideas to a trusted source at an healthcare client. During the discussions, we came up with few more potential applications. And in the process ironically I landed up with a request not to publish further details. Hopefully, at a future date I can share them with the world. :-(

Forward Looking Projects

In last couple of years, working as a freelance, my clients sometime float very interesting forward looking projects and ideas. These ideas sometime start out sounding wacky. But I enjoy the challenge to think, brainstorm and work with colleagues with totally different backgrounds across organizational silos. In the end, I usually land up researching feasibility of these projects for the clients anyway.

I wonder if this blog could be used to start discussing such projects, sharing what my research revealed and getting feedback from others.

I mentioned one such project in my previous blog entry about a very large multi-petabyte archive starting at 2PB and growing to 10PB within three years or so.

During my research in finding the potential issues with creating such large archive, I came across several initiatives. These projects are focusing on addressing different issues: power consumption, heat generation, single global namespace, ability to search the archive for specific information and retrieve a specific file or document within reasonable time etc.

Internet Archive seems to be the right place for starting such work as they are attempting to archive a large portion of Internet. I am particularly interested in the work, CR Saikley and team is doing on PetaBox at Capricorn Technologies focusing on power consumption and heat dissipation. Another interesting project is being done by Aloke Guha and his team on power management and MAID at Copan Systems.

In my opinion, there are quite a few other issues that need to be addressed before a reasonable solution for a large multi-petabyte archive can become a reality.

Same Seat Assigned to Two Passengers and Airline Chase to Bottom

Last week, I traveled on business to Phoenix, AZ using America West Airlines. Interestingly, the airline assigned several passengers same seat on the plane. It is just not the case of overbooking as these passengers were issued boarding passes with same seat numbers printed on them while there were at least half-a-dozen seats empty. I don’t know enough about airline reservation and seat assignment system. I am assuming the seat assignment is database driven. And found it strange that a seat number which is a unique with in a plane got assigned on multiple boarding passes … Is it a quirk in the seat assignment system or database design problem?

Also, what’s up with all airlines following Southwest and JetBlue business model in trying to become “me too” no-frills airlines and ignoring premium category? It looks like a “sheep herd” mentality of the airlines. I wonder when a startup airline will come around taking advantage of a vacuum being created by these airlines in premium travel space.

Continuous Data Protection & Availability (CDPA) 30,000 Feet View

The illustration is my attempt to describe the process for block level CDPA with 30,000 feet view. The concept is similar to versioning & journaling and database transaction logs. The main differences from other CDPA solutions are database driven CDPA tracker and removing the time-slice concept.



I believe a combination of file and block level CDPA can achieve optimal level of continuous data protection (CDP) desired by users. I hope CDP vendors wouldn't muck up the segment by positioning superiority of one level over another like virtualization or SAN vs. NAS.

It is not amusing anymore explaining that SAN and NAS are complementary technologies not 'competing' solutions.

Anyway, enough about CDP, it is time to get back to developing requirement definition for a large archive (2PB to 10PB) and creating a backup tutorial.