Last September, I wrote an extended post on "Overcoming Distance".
In it, I made the case that real-world issues around coordinating multiple copies of information at significant distances permeated so much of how we think about IT on a global scale.
Now, looking back, I realize my mindset might have been wrong.
Perhaps we need to start thinking in terms of exploiting what distance can do for us, rather than simply overcoming it.
Origins
If you read the post, you'll realize that it was triggered by the joint work EMC, Cisco and VMware did around an extended distance VMotion solution last summer.
Interesting note: many vendors have gotten sloppy and started to use the term "long distance" to apply to synchronous distances, i.e. < 200km. Once real-world considerations apply (network distance vs. geographical distance), you're often looking at 100km separation or less.
Esteemed colleagues, replicating from San Francisco to San Jose is not "long distance". Replicating from Singapore to NYC is "long distance". What you're describing, folks, is either "metro" or "extended" distances.
Just so ya know ...
A Thought Exercise
Imagine, just for a moment, that we were able to place data where it needed to be, and didn't have to worry about things like latency and bandwidth. Or the costs associated with multiple copies of data. Or, the real prize -- not having to worry about consistency and coherency between remote copies.
Perhaps the most powerful analogy would be quantum entanglement -- two particles that share properties, independent of distance. If your physics education stopped with Einstein relativity, the quantum world is a fascinating subject.
So, image we now have moved beyond ordinary electrons and magnetic media, and had the ability to (mostly) create the illusion of the right information, in the right place, at the right time -- and at the right cost.
How would our thinking change?
New Forms Of Protection
When information has a defined location (i.e. an array in a data center, your laptop, etc.), a key aspect of protecting it is making a copy at a distance. The more distance, the better. The more copies, the better. Hardware fails. Software fails. People make mistakes. Bad things can happen to physical objects, including storage devices.
An extreme example of this would be comparing, for example, a local copy on the same physical storage device to, say, 14 redundant copies scattered redundantly in data centers around the globe. If doing the latter was as cheap and easy and secure as doing the former, why wouldn't you?
So much of our thinking around data protection (at least for larger enterprises) involves notions of distance. People understand that distance matter -- more distance, more copies, more protection. However, there's an interesting juggling act in understanding the costs and benefits of this sort of insurance.
However, if insurance became "free" (or dramatically reduced in cost), would we all insure our information far better? Would we think of entirely new data protection strategies?
And -- more importantly -- how long would it take our thinking to collectively change?
New Forms Of Cost Optimization
I think that everyone understands the benefits of resource pooling. Whether it's servers, storage and network -- shared/pooled consumption models are generally more efficient (and responsive!) than dedicated resources.
Pool stuff in a small cluster with VMware -- that's good. Pool more stuff in a larger data center cluster using VMotion -- even better. Pool across multiple data centers -- much, much better. And dynamically pool across compatible service providers -- wherever they might be -- we're pretty close to pooling nirvana.
It's one thing to move a compute image around. It's another thing to move terabytes of information, keep it consistent, move it predictably and cost-effectively, and so on.
But, you'll have to admit, in this imaginary world I'm describing, we'd certainly have an entirely different take on cost optimization -- not only physical resources, but labor resources as well.
In this world, we'd think very differently about how many data centers we'd need, how big they were, and how they related to other data centers -- whether owned or provided as a service.
How long would it take our thinking to collectively change?
New Forms Of User Experiences
From an end user perspective, data at a distance sucks. The more data, the more distance, the more it sucks. Getting more of the data closer to the user results in a better user experience, which means that the user can do more with the data.
It's that simple, really.
In our proposed world, let's imagine that information can be dynamically relocated closer to the user at the right time (before you need it!) and the right cost.
How would this change how we build applications? How we think of IT infrastructure? How long would it take our thinking to collectively change?
Can Technology Outpace Our Ability To Re-Think Basic Assumptions?
I am starting to assert that the rate of technological change is starting to exceed our ability to assimilate its impact and leverage it effectively. The stuff seems to be available quite a while before we realize what's now possible, and how to put it into practice.
As an example, Google Apps (and similar offerings) have been around for quite a while. But it seems to take quite a while for a few IT organizations to come to the conclusion that -- yes -- maybe there's another way of doing desktop productivity.
By way of comparison, my college-age daughter prefers online apps to desktop resident apps. For one thing, all she has to do is get to a browser -- any browser -- and her stuff is there, ready to share. She feels traditional desktop apps are about as fun as hexadecimal code.
I think virtualization (and specifically VMware) is another excellent example. The art of what is now possible seems to lag decent adoption by an increasingly long length of time.
And as the technologists come up with new and wonderful ways to change our basic assumptions around how IT is done, will we continue to blame the technologists for slow adoption -- or start to blame ourselves?
Food for thought ...
Comments