Still think the cloud (or should that be The Cloud?)is a possible-but-not-definite trend? Take a look at IDC’s projection of IT deployment types:
So much to unpack! What really jumps out is that
- Traditional data centers drop in share, but hang in there around 50%: self-managed hardware will be a fact of life as far out as we can see
- Public cloud will double by 2021, but it isn’t devouring everything, because in the final analysis no Operations team wants to give up all control
- Private cloud expands rapidly, as the skills to use the technology become more widespread
- But most importantly…in the very near future, most every shop will likely be running a hybrid network, which combines traditional data centers, private cloud deployments, public clouds for storage and computation, and CDNs for delivery (don’t forget that Cisco famously predicted over half of all Internet traffic would traverse a CDN by the year after next)
It’s a brave new world, indeed, that has so many options in it.
If it is true, though, that cloud computing will be a $162B a year business by 2020 (per Gartner), and that 74% of technology CFOs say cloud computing will have the most measurable impact on their business in 2017, that means this year will end up having been one of upheaval, and of transformation. As ever more complex permutations of public/private infrastructure hit the market, the challenges of keeping everything straight will rapidly multiply: can one truly be said to be optimizing if one cannot centralize the tracking and traffic management for all resources, regardless of whether they’re in your own NOC, under Amazon’s tender care in Virginia, or located at some unidentified POP somewhere in Western Europe?
The truth is that, as with all transformations, this move to hybrid networks will be marked by the classic Hype Cycle:
We are fast approaching the Peak of Inflated Expectations; the sudden fall into the Trough of Disillusionment will be precipitated by the realization that there are now so many different sources of computation in the mix that nobody is quite sure where the savings are. Perhaps we’re saving money by using different CDNs in different geographies – but it’s hard to tell if we’re balancing for economic benefit; perhaps we’re making the right move by storing all our images on a global cloud, but it’s hard to tell whether adding a second (with the inevitable growth in storage fees) would result in faster audience growth; perhaps we’re right to avoid sending content requests back to origin, but at the same time, that seems like a lot of resources to not use.
The Slope of Enlightenment will hit when the tools come along to put all the metrics of all the elements of the hybrid network onto a single pane: balancing between nodes that are, at an abstract level at least, equally measurable, configurable, and tunable will start us down the path to the Plateau of Productivity.
The Cloud is coming; how long we spend in the Trough of Disillusionment trying to figure out how to make it hum like a well-oiled machine is assuredly on us.