kplaxmaster is a good friend and a bit of a rival blogger in the technology space. He wrote an article expanding his position from a recent argument on Twitter. I encourage you to go read it and formulate your own opinions before reading my response. Did you go? Great. So can you agree with me that the article starts with “cloud sucks” and ends with “WHERE ARE MY ELECTRIC CARS?!?”, effectively emulating the maniacal rantings of a backwater technomage stuck on dial-up. Just teasing, of course.
Lets start with the definition, which I believe hits the core of my disagreement. It seems that accuracy gets lost in reducing the cloud offering to “multiple computers act as one“. I’d much prefer that the community agrees to “cloud computing means (1)accessing/consuming/manipulating data on (2)any device (3)regardless of origin“. And I can provide examples of these interactions:
(1)I access my email through Gmail, I consume films from Netflix, and I make changes to documents in Google Docs.
(2) I do all of these things from my Android phone as I ride to work on the train for mobility. I do all of these things from either of my two personal laptops or my work pc (except Netflix) if I’m on the road or in the office. I do all of these things from my 8-Core, 3 Monitor home computer to get high-speed and simultaneous processing. I consume data on my Roku (connected to my TV).
(3)I can create a document and type an outline on my phone, when I get home I finish typing each of the devices. When I start watching an episode of Star Trek TNG on my PC and have to jump on the train to see friends I just open Netflix on my way out of the door and resume playing the same episode right where Q left off.
I think his statment about computation power costing more is extremely dubious: “This will drive down cost of PCs, yes, but also put even more trust into the cloud—and most likely raise the price of computers with actual CPU power such as a quad core 3.0 GHz computer as demand will go down.” This assumes that entire swaths of the CPU market is content to slow their inventive nature in favor of ‘churning out’ CPUs with minor improvements. It also assumes that server-side CPU developments will not impact costs to the end user. The market can centralize CPU processing, but that doesn’t reduce peak-performance requirements as processing is still taking place. I would even hypothesize that centralization drives computing requirements UP as server owners will be looking for any efficiencies they can find to reduce server footprint and compute costs during computation cycles that surpass.
I completely agree with the assessment that processing and bandwidth should be considered like natural resources just like electricity. I argue that such pricing has been grossly absent from the entire discussion, just look at the fact that American consumer pays for “bandwidth” instead of “throughput”. Even though I hate to say it, the wireless market is on the right track…just with the wrong pricing models in mind. They view 1GB of data as a fixed price when the reality is far more like the electricity delivery model: peak usage translates to peak pricing. If I’m trying to download a video in the evenings, during peak hours, it should cost more and I (and the service provider) should make decisions based on hard figures. Fixed lines (e.g. FIOS) have high up-front costs and limited channel sharing so consumers get truly dedicated lines, but we all dump into the same “distribution system”, to borrow a power term. Unfortunately, the similarities only go so-far as throughput isn’t something you can save up like electricity to be expended by another consumer. Regardless, we need to decide what we’re selling to users.
Now, your guess is as good as mine about why he starts freaking out about not having MS Kinect hooked to traffic lights or computers in refridgerators. But the question about why we need everything connected to te cloud is easy: because machine-to-machine management is hard. The average end user doesn’t want to have to understand client-server architectures, install UPSs in their homes for server PCs and reset devices to get them to resync. They also don’t want to debug misconfigurations, diagnose problems, or defend the contents of their refridgerator against cyber attacks from the server they forgot to power cycle after the last Microsoft patch.
But back to the core idea of trade off’s, it’s important to remember that a technology user is balancing three costs:
- [re]education – How do I do this again? I’ve never heard of that before!
- time – After I brush my teeth, it’s going to take me 5 minutes to hook my device to my PC and find the album I wanted to download before I hop on the bus.
- stress – The service is down?! Do I have to reset the server in the basement? Am I going to miss that call with Jane tonight because of some problem with MY stuff?
- money – Derived from the North American ‘evilroot’
The fact is that cloud offerings (right now), offer about as close to an “on-off” switch as you can get for education, 6 second buffering times for Netflix on 3G, no system maintenence stress, and for a price that’s well below most of the traditional service offerings (often free, once you assume connectivity). Now, you can claim that connectivity has to factor into the pricing, which I’m okay with. I personally pay $70 a month for connectivity ($30 cell data + $40 home internet). Add $17 for media and now I’m streaming years of content (video and music), communicating on every social network and through commercial channels for $97 dollars per month.
You can argue that those 6 seconds of buffering and $30 dollars per month would be justified by having to sync all my devices at home on my PC before I leave for work, but the time to download ALL of my media plus not having the ability to choose on the fly means I’m not ever going that way. And neither are the data providers: it’s simply TOO HARD to manage DRM in a distributed computing system. So if I don’t want to move and they don’t want to move…why resist?
Filed under technology | No Comments »