[kictanet] [KICTAnet] TEAMS | EASSY Fiber Cables Cut? SEACOM | LION?

[Michuki Mwangi]

On 2/29/12 9:56 AM, Brian Munyao Longwe wrote:
> Dear Robert,
> 
> Like it or not Nicholas' point is fundamentally true.
> 
> Let's go back to Internet 101: You either have eyeballs or you have
> content. If you have content, the eyeballs will need to get to you so
> you need to sit your content somewhere accessible, affordable and
> secure. If you have eyeballs you need to provide them with a means to
> get to the content or else they will go to someone else who can.
>

+ 1

> The problem with Kenya is not just that our energy is expensive. The
> biggest problem is that it simply is not enough. Ask Dr. Ndemo or Paul
> Kukubo what one of the leading questions any content owner asks when
> they come visiting to find out what all the buzz is about Kenya and
> whether they should consider putting up here? Yes, you got it - "can
> you meet our energy needs?" Without a fail in every case the answer is
> "no".
> 

+1

> This is an issue that needs to be considered very carefully if we
> intend to be a serious contender on the global scene.
> 

Am not particularly in agreement with this statement in view of the
comments above. I will substantiate this below.

> Tried to find online stats about the average energy usage of a Google
> data centet but apparently these are well kept secrets. But I remember
> hearing sometime back that a single Google data center uses more
> energy than the city of Nairobi(sic).
> 

I have information on Equinix and their largest data centers consume
80MW of power. (FYI Equinix is the largest Carrier Neutral Data Center
provider in the world).


While i appreciate our desire to be a player in the international scene
and host large content providers, etc. We tend to forget that when
Google, et al are building Data centers their objective is to serve
several hundred million users from those locations.

IMHO, we need to rethink the business case for hosting. I believe it is
under-developed and needs more attention. Let me try and use what we
already know to explain this.

Q. According to CCK there are 14.3 Million Internet Users in Kenya so if
we are building hosting infrastructure (power being the main concern)
how much power do we need to host content for 20 million users?.

A. We have sufficient capacity today (power and colo space in Kenya) to
host content for 20 Million users. Fact - Google Cache's in kenya serve
at least 20% of most ISPs traffic. The foot print of these servers
(space and power) is tiny.

Q. To further expound on this, what foot print is required to run a data
center providing a local Free Webmail service targeting 20 Million users?.

A. Too small with today's computing resources and green techniques.

Am actually curious to know what percentage of data centers in Kenya are
running at least 50% of their installed capacity in relation to their
operating costs. By this i mean, are their costs indicative of operating
at 100% or at the 50% especially power, cooling, tech support, etc.


In conclusion, am more convinced that we need to start developing
hosting solutions to serve Kenya before we can serve the world. Right
now we are aiming to serve the world before we can serve Kenya. This is
very evident with the fact that a server aimed at providing KCSE results
is unable to support queries at peak times, but on the other hand we are
keen on getting content owners to host locally.

It is common practice by investors to make unrealistic demands as a
polite way to put off weak proposals. So maybe we should look at the
power requirements from that point of view.


Regards,

Michuki.