Monday, April 25, 2011

Forecast: Cloudy With a Chance of Outages

I used to think I understood copyright law.  The laws were straightforward and, for the most part, reasonable, and the fair use provisions clear and unambiguous.  But of course, over the last couple decades, the copyright holders of Big Content have had little difficulty convincing the courts to support an increasingly arbitrary and even draconian interpretation of those laws.  One of the shining examples of this willful manipulation of the laws governing a consumer's use of legitimate IP is the case of UMG vs., where the determination was made that you could legally purchase a song, and you could legally purchase some online storage where you could keep your files, but you could not legally store that song in that online storage.  That, you see, amounted to piracy.  Absurd.

But absurd though it was, it became the conventional wisdom that the concept of a cloud-based file "locker" for a consumer to keep his or her music, so that it might be accessible from any device anywhere, was dead on arrival without a license and the accompanying payments to every major record label.  And while the labels have gotten on board with streaming services and even digital distributors, there has been no path to a legitimate online storage system that could earn their approval.  Of course, MP3 files are just another collection of bits, so consumers have been storing their music in the cloud if they wished for years, but obviously an approved system could be optimized for that purpose.

Which brings us to Amazon.  Clearly, they viewed the court's decision in UMG v as silly and indefensible, so they announced their Cloud Drive music locker service and Cloud Player software that allowed you to stream your music from the Cloud Drive to most any internet connected device.  The service has been up and running for weeks now, and so far all we've heard from the ordinarily highly litigious record labels has been some muttered condemnation and a bunch of crickets.  Could it be that they don't think they could win this argument in court?   Yeah, I think so too.  To ask a Federal judge in 2011 to prevent people from using online storage for their music library would be significantly over the top, and a loss in court would announce to the world that an online music locker service was a business model that could operate unfettered by music industry legal harassment.

Both Apple and Google have been planning to roll out a similar service, but they have been restrained by their belief that they would have to win approval from and pay licensing fees to the copyright owners before it could be deployed.  Now, they are both watching closely to see what the response of the industry will be to Amazon's service.  And the labels are faced with a real quandary.  If they don't challenge Amazon in court, at some point both Apple and Google, and probably a number of others, will decide that there is no real legal risk in providing a similar service, and the genie will be well and truly out of the bottle.  But if the labels DO sue Amazon and lose, then they have no basis for threatening at least the smaller players - really, they'd have no potential revenue stream at all from cloud storage.

Which, in the end, offers the only truly valuable insight from the whole ugly rear-guard fight the music industry has put up against technological change.  The labels should have owned the whole market.  They should have been iTunes, Rhapsody, Amazon - the whole digital distribution ecosystem, just as they have always owned all phases of physical distribution.  They could have led the way on R&D, partnered with electronics companies to build music players, built data centers - in short, they could have leveraged the technology to increase their revenues and control, rather than see all those pieces go into different, non-traditional hands while the record companies find themselves with an ever-shrinking piece of the pie.

Of course, the cloud comes with its own special challenges.  Last week a problem with Amazon's EC2 cloud compute platform knocked a whole bunch of web - based businesses off line.  The explanation is that an error caused a great deal of the associated storage to start to re-mirror, and this used up all the processor cycles and all the available storage and bandwidth in that zone.  So everything stopped.  Of course, the non-technical press (and to be fair, a decent portion of the technical press) ran stories with headlines like "The Cloud: Not Ready for Prime Time?" and "Is The Cloud Just Too Risky For Your Business?", without the slightest sense that these are not only stupid questions, they represent a category error of the first order.

Two points have to be made about utilizing a cloud infrastructure.  First, it's absolutely necessary to understand that this is an economic decision, rather than a technical one.  That is, there is a minimum baseline amount of IT infrastructure necessary to run the core business.  Beyond that, there is development, new features, surges in demand, special events, any number of things that can bring that baseline infrastructure to its knees.  Today, you just add fifty or a hundred servers in the cloud and manage demand and resources dynamically.   Obviously, in the past, businesses didn't have, and couldn't justify procuring, those hundred servers, so the site just crashed anyway.  The cloud changes the equation, and makes it easier for smart people to deploy innovative new services, because the barrier to entry isn't buying and maintaining five hundred or a thousand computers anymore.

Second, if you put your business in the cloud, the cloud is your datacenter.  Treat it as such.  Redundancy, multiple points of failure, failover

But no matter how many of these silly scare stories about the risks of cloud computing infrastructure you hear in the coming months, you need to understand just how stupid, pointless and without meaning they truly are.  The web already is entirely dependent upon cloud based compute and storage capacity, and it would be economically impossible to return to the days when all your server metal was in your building. This "compute elasticity" is actually a very good thing, both from the standpoint of allowing innovations on the web that let you do cool stuff you like to do, and from the standpoint of giving all of us consumers a better option for storing and accessing our digital stuff.

Ain't technology wonderful?


  1. Ain't technology wonderful?

    Just you wait until The Man begins replacing us all with robot clones, mikey.

    THEN we'll see...

  2. A problem with the cloud is privacy. It's pretty much against the law for my workplace to store off-site. The various clouds will still have national bases, and people in one country will have to decide if it's worth it to allow people in the next country to store all their stuff.

    For trivialities like mp3 libraries it's fine, but there are a lot of uses for which the cloud seems nearly dangerous.

  3. Actually, the security argument is utterly specious, a red herring. Where was Sony's data? Epsilon's? Card Service's? The fact is that data in the cloud can be significantly MORE secure than locally stored data, because, due to the economies of scale, they can employ more and better security practices and personnel. Most companies and organizations are good at OTHER things besides running a data center, and it makes sense to outsource the data center function to a company that specializes in running a highly available, secure, robust data center.

    It's important to remember that we haven't figured out how to protect data absolutely, so if security vulnerability is the defining decision-making methodology, you cannot use computers in networks at all - only in an air-gapped stand-alone mode. And if you aren't doing this (some government agencies actually DO) then you are already vulnerable and you need to make decisions based on a wider set of criteria.

    In one sense, that of data persistence/recoverability, the cloud is already superior to a local data center. They have much greater capacity and capability for robust backup/snapshot/restore/failover functionality than most companies could ever DREAM of.

    These sorts of arguments will fade away as the world becomes more comfortable adopting a more flexible, fast, affordable method for deploying compute and storage resources...

  4. They might, but privacy laws are moving forward everywhere but in the US. I can't see why the government of France is gonna trust any information - patient records for instance - to any other nation-based cloud. It was a very big deal here when medical records went outside the province, and should be as America is known for intrusiveness.