The sad state of the cluster – project on hold

Sema kitu uone.

And so the blog post came. The iHub cluster is alive. She purrs. We are going to do great things with the awesome power of the nodes. It was meant to be, initially, a research project that would, eventually have a self-sustenance model.

And the revenue potential for the cluster was immense. Given time and support.

Google MFWIC, Eric Schmidt came in and he said: “This is exactly how we do our clusters”. I was flattered. I have built, by hand, in 4 days, Africa’s first Ati-GPU SuperComputer cluster north of Limpompo.

But then that is as much as it amounted to.

The next day, I was called to a meeting.

And this was the news:

“Salim, the funding that Google gave for the cluster has gotten depleted. We can no longer cover the cluster’s admin costs and unless you will work for free, we are putting the project on hold. Also, unless it is profitable by March, 2013, it will be shut down.”

I was inundated in utter melancholy. WHY!!??

So, as far as I am concerned, the project has been halted. Not for lack of resources. Ohh no. Not for lack of potential. No. But for what I see as clear lack of goodwill and BELIEF that the cluster will amount to something.

I am described as opinionated. Pithy. Blunt. So here we go again.

I will share the Locus Standi, and if it seems like Salim (The non-conformist rebel) is ranting, then please stop reading, so that you don’t do a Mbwana on me and refuse to share a lift with me. Ahem!

The cluster was my idea. The working super-computer was the work of my hands. I did the proposal and got the funding from google. Also, I would like to believe that the idea of the “cluster” added weight to the joint-proposal sent to Google by iHub for funding.

Now, I have no problem working for free. I have been doing this for a great part of my life. I have lived a life where I do all the work, and other people get all the plaudits and all the money. So, working for free was not the issue.

My biggest beef is that this will unfairly go down as another project that Salim “has started but not completed”. And this does not let me sleep well at night. Especially given the potential of the cluster.

So, I started asking questions. I demanded to see the accounts. How has USD 50, 000 ended just like that.

As per the logs and calculations:

  • Rent for 1 year (Sept 2012 to Sept 2013) – USD 10k – As advised during the budget phase.
  • Equipment for 4 nodes – USD 14, 000 (Later revised as 20, 000)
  • Multiflex server for head node – Donated by Intel (Free)
  • 2 people pay (July, Aug, Sept), One person Oct pay USD 1500 per month – USD 10, 500
  • Server Rack and Cabling – USD 10k

Total : 44, 500 (or 50, 000)

So, clearly, the cluster needed more than the 50k. Good no problem. The allocated money was depleted. And all was accountable.

But I started asking more questions.

With all due respect, why was the cluster given the SMALLEST portion of the Google’s 250k. Why was UX Lab been given USD 100k, and the SuperComputer 50k? We are short of hardware and money to “operate”, whereas the financially-doped UX Labs changes the floor depending on the length of delays of Mpesa transactions and the color of Uchumi mandazis. This week Ceramic. Next week Marble. Next week Pink Tiles.

Is the focus here technology, or pedicure? Does Africa need a setup to tell developers what hue of Hazel to use on their app icons more than they need a super-computer?

I always say, Never blame the car for not taking you to your desired destination, if you use the fuel money for cologne and ceramic floors.

For some reason, there is this annoying premise that since Salim can do it, any off-the-streets campus kid can do it, with a little googling.

And it is for this reason that, I cannot say I was the least impressed when I went to the cluster room on Thursday (Feb 7), barely a month after I worked day and night to build it  and found an intern having formatted the MASTER node of the cluster for “leaning purposes”. Without as much as a phonecall to consult with me. God bless the intern. The cluster is dead.

Page 1 of 2 | Next page