One of the starkest ironies of life is efficiency. xkcd I believe, has posted some comics on this issue, a prominent one being http://www.explainxkcd.com/wiki/index.php/1445:_Efficiency
There is a cost to optimization. If we're trying to make one workflow efficient, optimize some measurable goal, then there's a whole plethora of costs that we willingly, unwillingly or unknowingly compromise on. Compromises in personal life vs professional life have been quite well explored in popular film media, but I think the less sexy issues aren't.
Lets take for example your personal computer. It's a popular meme that the PCs of today are far more powerful than the computers combined that helped send man to the moon. Lets think about it a little more. At my workplace we have 70 people, each using at least one Macbook Pro. This is a PC with 8 gigs of RAM, 500 GB memory many GHz (Or gflops if you like) of processing capability. Most of the time the processor remains at near idle rate. Most of the people are using the MacBook for Office applications like Text editing (which includes coding), making presentations, browsing, communicating and watching/reading stuff. Sometimes there's more processor intensive applications like rendering, video editing, designing and compiling code.
If you combine the requirements of everyone, you could maybe take 10 Macbooks, put them up as a cloud server, give everyone a simple terminal and even then you would not need more power even during high load times. What's a more striking number is that the total Hard disk storage used by everyone combined is not much more than these 10 PCs together.
Why then do we not have this system?
It's all about compromise. The jobs we perform are really complicated and require complicated machines which in turn means that there are a lot of avenues for failure. If there's some problem with the cluster, then suddenly most people's productivity dies. With distributed PCs, you have load capability at the same time as reducing risk.
But that is not to say that there are no avenues for cost minimization. If you have a big enough firm, one can still do a cost analysis to find the point at which a well setup cloud cluster, with enough backup capability and redundancy is more cost effective than personal laptops. There is still a tonne of avenues for innovation in the world. This is exciting.
There is a cost to optimization. If we're trying to make one workflow efficient, optimize some measurable goal, then there's a whole plethora of costs that we willingly, unwillingly or unknowingly compromise on. Compromises in personal life vs professional life have been quite well explored in popular film media, but I think the less sexy issues aren't.
Lets take for example your personal computer. It's a popular meme that the PCs of today are far more powerful than the computers combined that helped send man to the moon. Lets think about it a little more. At my workplace we have 70 people, each using at least one Macbook Pro. This is a PC with 8 gigs of RAM, 500 GB memory many GHz (Or gflops if you like) of processing capability. Most of the time the processor remains at near idle rate. Most of the people are using the MacBook for Office applications like Text editing (which includes coding), making presentations, browsing, communicating and watching/reading stuff. Sometimes there's more processor intensive applications like rendering, video editing, designing and compiling code.
If you combine the requirements of everyone, you could maybe take 10 Macbooks, put them up as a cloud server, give everyone a simple terminal and even then you would not need more power even during high load times. What's a more striking number is that the total Hard disk storage used by everyone combined is not much more than these 10 PCs together.
Why then do we not have this system?
It's all about compromise. The jobs we perform are really complicated and require complicated machines which in turn means that there are a lot of avenues for failure. If there's some problem with the cluster, then suddenly most people's productivity dies. With distributed PCs, you have load capability at the same time as reducing risk.
But that is not to say that there are no avenues for cost minimization. If you have a big enough firm, one can still do a cost analysis to find the point at which a well setup cloud cluster, with enough backup capability and redundancy is more cost effective than personal laptops. There is still a tonne of avenues for innovation in the world. This is exciting.
No comments:
Post a Comment