Software Project Success – Prototype First

November 13, 2012

As I review the successful (defined as on-time and within budget)  & unsuccessful projects that I have led over the past 10 years, a common thread has been that successful projects had at least 80% complete prototype prior to start of any development work.  This is spread across both Agile and waterfall methodologies.

The prototype allowed the Business Analysts to “touch” the product and better visualize & communicate what they expect of the product.

To get a project kicked off, start with a UI Developer and a Business Analyst.  Refine requirements till you are 80% complete and then bring Software Architects.

This may seem contrary to several Agile approaches, but in my perspective it is a modified Agile approach that has worked for my teams.  The agility is divided into 2 phases – Agile requirement gathering (with UI developer & business Analyst) and then move to Agile software development.

I have also noticed that bringing software developers into the process prior to the product vision being ready,  tends to compromise the product vision and it is best to involve them in architecture & development once the vision is clear.

Advertisements

Talent availability important factor in technology choice

November 11, 2012

I am mentoring  a startup that chose RoR because the founding team liked it.  However, they are unable to find local talent sufficiently interested in an entrepreneurial venture in the Houston area.  Sure, one can find remote resources, but for startups I am a strong believer that the technical co-founder should be in same city as the founder.  It is definitely OK and probably even better if rest of the team are at offshore or remote sites.

Availability of talent is a critical component of technology choice!

 


Scared of eDiscovery?

November 11, 2012

Are you scared of eDiscovery?  You are not alone.

eDSG ran a one day poll asking “What’s So Scary about eDiscovery”.  756 Respondents provided the following results:

  • The Cost = 32.25%
  • All the New Technology = 17.5%
  • Giving up Control to Outside Counsel = 11.75%
  • Giving up Control to Third Party Vendors = 11.75%
  • Lack of Internal Support for IT = 9%
  • Lack of Technology   = 9%
  • Other Reason = 9%

Other Reasons Listed included:

  • Having to deal with Lawyers = 15 responses
  • Having to deal with eDiscovery Sales People = 35 responses
  • Self Collection = 18 responses

Move to the Cloud

November 11, 2012

I just completed a very thorough study evaluating costs of maintaining our internal IT infrastructure or moving to the Cloud. My conclusion is that the cloud is compelling!!
The ability to start small retaining higher margins, no forward planning required on hardware infrastructure, lowered Total cost of ownership and more importantly the ability for rapid deployment thus supporting new Business models make moving to the cloud very obvious.

 


Justifying Move to the cloud

November 9, 2012

Excellent articles explaining ROI to move traditional IT infrastructure to the Cloud

http://goo.gl/oFyhj

http://goo.gl/4eNTd


Technology Startup without Technical co-founders – Ridiculous

November 9, 2012

It is amazing to see a large number of technology companies without a hands-on technical co-founder.  On face value, it appears you could outsource the job  or provide equity to an organization. I have however yet to see such a successful arrangement.

For non-technical founders, I recommend you bite the bullet and take time to find  the right technical co-founder.  It will make a difference to your startup!!


Corona to the rescue of Hadoop

November 9, 2012

Batch processing jobs still not meeting user expectations after putting Hadoop in the mix?  Great article with a good analogy in regards to bottlenecks while grocery shopping

http://tinyurl.com/b4kzemk

Corona divides the job tracker’s responsibilities in two. First, a new manager manages cluster resources and keeps an eye on what’s available in that cluster. At the same time, Corona creates a dedicated job tracker for each job, which means the job tracker no longer has to be tied to the cluster. With Corona, smaller jobs can be processed right on the requester’s own machine.

Will this help improve overall throughput?  Looking forward to giving this a dry-run.