Of mice and migrations: How a rodent's DNA maps to architectural complexity Managers of enterprise systems are being bombarded by messages touting the supposed benefits of cloud for cost reductions and greater IT “flexibility.” Bosses are pressured to, for instance, develop a hybrid solution of off- and on-premises systems: the difficult stuff is left in the company’s private cloud, and the rest …
"The demand placed on IT is that we attempt to hack systems more complex than the genome of a mouse and do so blind, doing it randomly and with minimum expenditure.
The hope is you’ll obtain a hamster, but don’t be surprised if you get a rat that bits you in the apps."
I do think that that is a brilliant summary of the state of IT today.
...really?
A group of humans built that IT system. I think it might be a while before a similarly-sized group of humans manages to build a complex, multi-cellular organism like a mouse. There's probably one or two lines missing from your mouse network diagram, plus that whole real-time, massively-parallel, multiple-sensor to consider.
Nice analogy, but don't over-sell yourself.
This is what one thinks when one knows a bit about genetics but nothing about biochemistry. I remember being in exactly that position (having got a degree in the former, and starting a PhD in the latter) and expressing to my supervisor that I was still amazed that DNA could make a mouse.
I've never forgotten his reply. DNA, for all it's complexity, is just data. You can put a mouse genome on a CD; you can mail (or even email) it to people; you can do all sorts of analysis on it. But the only thing that can turn it into a mouse is the molecular and biochemical machinery in a mouse embryo.
@Andy
Leaving aside for a second that its an analogy and by definition imperfect - are you really so sure of your statement? Firstly lets assume each one of those nodes is an IT product - either software or hardware it doesnt matter - then the number of people involved must include those that built that component - so we're easily hitting >million people already ie assume an Oracle, a Microsoft and IBM and an Intel are all involved.
Secondly the DNA is just the blueprint (TDD if you like) - most of the complexity of the mouse is not expressed directly by the DNA but by the Proteins produced by DNA. Look up Proteome.
So fail on both levels really.
Well, just a network map of DNA on its own would be blank. You need the expression and the homeoboxes and the epigenetic stuff in order to start building a network up, so I think there must be some of that going on in there.
If those are simply genetic interactions within a single cell, then the comparable object is the circuit diagram of a PC, not an internetwork of PCs.
> A group of humans built that IT system. I think it might be a while before a similarly-sized group of humans manages to build a complex, multi-cellular organism like a mouse.
The IT system wasn't built, it was more likely thrown together more or less randomly, evolving over time according to the demands of the business. The mouse genome, on the other hand...
If you left it up to mice to build a computer, then it would probably come up with the answer 42 and require an even bigger computer to be built in order to understand what question produces that answer.
A fully connected graph is roughly n(n-1) cross connections, which is a lot of links.
But the basic point, that interaction are more numerous than most people would expect at first glance and doing something like chopping a whole wodge of them out and putting them into an off site cloud is asking for lots of trouble, stands.
TL:DR. Danger. Here be big f**king monsters.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
