Being a former Financial Service Information Technology executive I am often asked by startups; how do I position startups in large enterprise engagements. The answer is simple – have something that is of ardent interest to the organization that complements their existing infrastructure and skills. Keep in mind the most expensive aspect of any risk based enterprise venture is the cost of people, and loss of time, therefore your product/service must synergize with the existing architecture and development / support philosophy.
One such endeavor I recall that was a very successful Grid engagement that was lead approximately two years back within a major insurance company in NJ. You wouldn’t typically associate insurance companies with cutting edge technology; however this one was very proficient with multi-million dollar Massively Parallel Processing (MPP) platforms. Enter startup Topspin (TS) networks, which did their homework identifying major accounts and noted risk taking individuals engaging in this space of MPP technology. Essentially, TS switch created a virtual Grid architected exactly like IBM’s SP2 MPP switch, only TS switch was focused x86 nodes costing $1,500 each, opposed to IBM’s AIX nodes costing $250,000 at that time. The IBM MPP architecture was quickly mimicked with TS Infiniband switch and cost savings ranged from 10 to 15 cents on the dollar as compared to price performance on $MM MPP platforms.
Being an enterprise early adopter of Grid technology I was asked to present my findings at several different industry events -- displaying results and relative costs (use search engines for details). While speaking Stanford Networking Research Center (SNRC), I requested that Topspin and Venture Capitalist community recognize Microsoft as a key player in the enterprise space, and that any Grid platfrom would only realize fully enterprise utility when provisioning Microsoft OS is achieved. Well, my request was recently answered on September 16 2005; Topspin announced support for Microsoft Windows 2003 with high performance Infiniband protocol drivers. From my perspective, Topspin’s approach to Grid computing is the most ‘pure definition’ of utility Grid computing in the industry, combining open architecture flexibility, with high performance x86 computing.
So what are some of the lesson learned?
- Must align with enterprise corporate philosophy.
- Must be easy to install & maintain with little impact on existing staff & skills.
- Must partner / evolve with other technology standards (i.e. MPP)
- Must be open to other vendors platforms (i.e. UDB, Oracle and MS SQL server)
- Must be cost effective.
- Must beat existing technology in price & performance
Icing on the success story, Topspin was recently acquired by Cisco May 2005, for their market leadership and penetration into enterprise datacenters.