Three Ten-Year-Trends That Are Driving IT Change

A trick I use to predict the future is to identify ten-year-trends. These are really big trends that take place over a decade or more. At the moment, there are three ten-year-trends that interest me, but before I get into that, let me give some historical examples to clarify what I’m talking about.

In 1981, there were only an elite and nerdy few of us who had computers on our own desks. (I built my first computer in 1977, at age fourteen, but I am … different.) There were more people with some kind of remote terminal, but the majority of business people had no personal access to computers at all. Then, in 1982, IBM introduced the PC, and by 1992, ten years later, pretty much everybody in business had a computer on their desk. The ten-year-trend in IT during that decade was “put a computer in front of everybody’s face.”

During the 1990s, the ten-year-trend was “wire all of those computers into one big network.” At the beginning of the nineties, many computers weren’t networked at all, and networks that did exist were often small, connecting just a handful of computers in nearby offices. By 2000, the Internet was ubiquitous. IBM was running TV ads with nuns on camels in the middle of the desert sending e-mails to each other.

When I see lists of “this year’s top trends,” they often feel small and short-sighted to me. Any trend that only lasts a year is too insignificant to be important. It’s probably mostly over by the time you spot it. The really interesting trends, the ones that drive big change, last many years—a decade or more.

The three ten-year-trends that I see in IT today are: 1) Cloud/Outsourced Computing, 2) Server Virtualization, and 3) Flash Memory.

The big question behind cloud computing (or outsourced computing) is whether a company should build or expand its own data center, or whether it should access computing resources remotely, over the Internet. People are already using lots of cloud/outsourced computing today for things like blogging and web training, but I expect it to move up-market over time. This won’t happen overnight, but ten years from now I believe that many medium-sized businesses and a few large enterprises will have cloud-outsourced pretty much all of their computing infrastructure. (The term “cloud computing” is confusing because so many definitions are floating around. I’ll give mine in an up-coming blog.)

For people who do build data centers (either because they ignore the cloud/outsourcing trend or because they provide cloud computing to others), server virtualization will radically change how those data centers are built. Server virtualization is at the center of the trend, but it pulls a boatload of other trends along in its wake, including blade farms, network virtualization, and (my favorite) storage virtualization. Data centers built in ten years will look radically different than they look today.

Flash memory is amazing stuff. It is the first technology in decades with the potential to create a pervasive new layer in the storage hierarchy. Many technologies over the years, like bubble memory, have been touted as “disk-drive killers,” and none of them ever panned out, but flash memory is emerging as an enormous force, and there is no question that storage systems will look radically different in ten years. I’m not predicting that flash memory will eliminate disks any time soon, certainly not in five years and probably not in ten years, but I do think that it will relegate disks to an increasingly tape-like roll as flash absorbs more and more of the I/O intensive loads that disks handle today.

These are the three ten-year-trends that I see driving change in IT, but I’d love to hear from readers if they think I’ve missed some, or that the ones I’ve chosen won’t drive as much change as I think they will.

Comments

[David Slik]

I agree wholeheartedly that Flash is an underestimated and under-appreciated fundamental change to the storage and memory hierarchy that is such a key part of all computing infrastructure. It's not just storage that will be affected — flash technologies will change everything from networking hardware to desktop computers.

We're currently in the "horseless-buggy" stage of flash technology deployment, focusing on making flash look like existing devices, putting it behind a SAS or Fibre Channel interface, etc, but the really exciting possibilities come when the falling costs make it a new fundamental building block for designing completely new architectures.

For example, by utilizing intelligent power management and data placement across a large flash die, the same principles that are being deployed for MAID can be used to reduce the power consumption and heat dissipation to allow die stacking and other three-dimensional processes which further increase density and reduce costs. Instead of doing RAID at the drive level, we'll be intelligently replicating data at the flash page level to increase both resiliency and performance..

The intelligent use of flash as an cache layer that sits between DRAM and disk will also drive richer interfaces between applications and storage, so that knowledge of usage and access patterns can be leveraged in the storage system to intelligently place and pre-fretch data within an enlarged storage hierarchy.

And the exciting part is that like most ten-year trends, the really interesting emergent results will be the ones that comes and surprises us.

[Ian]

Hi Dave - I think the issue with predicting these megatrends is that it's the application of the trend by new adopters that makes the trend important. Also, I think a trend needs to have real and lasting impact and be important in cultural and political terms in order be significant. Both of your examples fit these criteria - it's a negative example, but do you really think the world would face the same challenges as a result of the proliferation of extremist views if it weren't for the personal computer and internet?

On your three proposed trends – cloud computing/SaaS I can buy as being important at a macro-economic, political and social level – particularly when combined with some of the low-cost computing initiatives that are being piloted in developing countries today. This will give the technologically excluded access to applications that they could have only dreamed of affording on the desktop PCs they once dreamed of owning  The other two, I have to say, I fear are a result of you being too close to the subject to see the bigger picture, but I guess we’ll see how they play out over the coming years.

[Val Bercovici]

Well, I'm glad to be actively working on #1 and #3 then :)

A great post and great comment discussion here. What I also find intriguing is how these 3 trends feed off each other to accelerate adoption of all.

Clouds need data centers which need virtualization for efficiency, which in turn needs flash for high-performance I/O. Read this sentence backwards (loosely) and replace "need" with "enable". It still works.

There are also 3 invisible supporting catalysts here (multi-core CPU/GPU's, ethernet convergence and of course pervasive dedupe) but that's the topic for a whole other discussion :)

[Stephen Foskett]

Dave,

I was writing a comment/response, but it got too long so I put it up on my blog:
http://blog.fosketts.net/2009/03/11/ten-year-trend-mobility/

The short version: While your three picks are important, they're symptoms of the singular trend: Mobility. In ten years we will look back and chuckle at our contained in-place datacenters and applications as quaint.

Stephen Foskett

[Bob D]

Dave,
If your right (and I tend to think your at least close ) regarding the trend toward cloud computing/SAAS then the marketplace that storage vendors compete in will change dramatically over the next ten years.

There will be a smaller number of customers buying storage. Those customers will purchase huge amounts of storage and many medium-sized businesses won't buy storage at all.

What will those large customers want?

[Michael Hay]

Chuck, I look at all three and agree they will be of impact. In particular FLASH because of the fact that it is so pervasive in consumer applications like the iPhone, and the Cloud seem to ring true for me. However, in the spirit of your most recent post, I disagree that server virtualization is one of the top ten year things. I actually think that the retirement of the Baby Boomers and the rise of the Millennials is of far more impact because it brings multiple technologies to the forefront that will reshape work behavior. Imagine an internal Twitter like service, but then think about how this material can be admitted into a court case under the FRCP. Or imagine the impact of an entire generation who has figured out peer-to-peer decentralized trust. I suggest that many of the technologies you are describing are the result of their impact on our society, and that makes them worthy of more study. So do you have a different opinion? :)

[SL]

One thing I see that will and can last a decade long is FCoE. FCoE will revolutionize the way on how networking works, not only in Data Center but how every computer in each home will be able to connect to each other in the future. FCoE is a start of something new and exciting.

[Dan]

I'll suggest a 4th trend (or an additional opinion). Data Explosion.

Server Virtualization and Flash Memory means higher density of data which means smaller footprints and an explosion of data in the datacenter.

I was just reading about the Bill Gates backed LSST Telescope (which you guys should look into) that will be using a 3Gigapixel camera to record 30 TB of data per night, leading to a multiple 100PB databases over the ten years.

Each picture on my 15MP dSLR consumes 20MB and I save 3 local and 1 online copies.

Video is just getting started...

With 2TB Memory Sticks hitting the market this year, data explosion is going need new solutions.

[Peng]

2) Server Virtualization and 3) Flash Memory seem to like by-products of 1) Cloud/Outsourced Computing.

I suggest a new end-user device combining mobile/pda/notbook be another trend to compliment the Cloud/Outsource Computing trend.

[Rajdeep Sengupta]

Dave,
I agree on three trends and on a broader picture, one is enabler for the other as someone pointed out. I can also see two more interesting trends, which are also interlinked to this.
1. With data size increasing exponentially, so hierarchal storage and an automatic mechanism which will intelligently work between primary, secondary and tertiary storage just like they way the things work between memory, cache and primary disk will be an important part of storage.
2.Another interesting trend is green IT consciousness, in future there will be more innovative approach towards energy efficient systems

[Fred]

Hi Dave,

I think mobility and green computing will be driving forces in the industry. These will drive the technologies you identified.

[Alan Moore]

Hi Dave

I recently posted my counter cyclical views on cloud computing on CloudeAve, a pro-cloud website a freind of mine contributes to. What I found interesting was that nobody commented! I reckon cloud computing is a risk to the mere existance of corporate IT not part of it! But what it doesn't do is solve any real problems and I think it is at risk of screwing things up even more for businesses.

Check out - http://www.cloudave.com/link/cloud-computing-the-new-nirvana-or-just-the-latest-version

Keep up the good work Dave - love your gear!

[A. Charlatan]

Dave, in the theme of your earlier post. I would like to offer an alternative view.

Requirement and capability drive change, but consumer want (not corporate need) drives trends.

I believe that in both your examples you state business changes that are the result of consumer, not corporate drivers.

The rise of PCs gave way to better personal communication, entertainment and information access. Most people don't need computers, they want the capability they offer and that's what drives proliferation and therein trends. I believe the same applies to your Internet example.

So in that vain, don't consumers already have cloud computing. Any service that can be engineered can be found and serviced from anywhere by Internet users now. Surely that is the value of cloud computing. The decentralisation of assets (people) and the flexibility of connection and working. Not the reduction of local management requirements and the outsource of capability at the expense of flexibility and control. Therefore, this, the move to home / remote working must be the trend for cloud computing to be the capability and I (unfortunately) don't see this as having a compatibly with human behaviour regardless of its overriding technical, social and economic merit.

Also, server vitalisation is surely only a stop gap if cloud computing is your goal. For Cloud computing to provide something more than hosting, (in my mind) it must be based first on centralisation but then on application fluidity. I don't want visualised servers or storage, I want visualised application portals capable of offering me the functionality I need, independent of my connection mechanism and location and intelligent enough to tailor to my needs and preferences (application / resource gateways if you like). In my mind, platform vitalisation is only a capability to make proper use of overpowered platforms with underpowered applications. Doesn't this fly in the face of cloud computing?

Again, to me flash memory, is to trends only what floppies were to the rise of PCs, a capability at the whim of scientific discovery.

Trends surely have to drive and be driven by consumer want and in my very (very) humble opinion, none of yours do.

I vote for

The decline of PCs,
The conglomeration of technology vendors,
The rise of the independently informed thinker.

[tara]

Simplification.

[Alex]

Hi Dave
I wonder if storing secret documents will cause a lot of touble with the advent of cloud computing.So does it have a chance ?

[GregB]

Hi Dave,
I keep wondering if several of your ten year trends are really just pendulum swings within a much longer cyclical trend.

Twenty-some years ago, I did my work on an operating system named "VM", which ran multiple operating system copies on a really big centralized server (mainframe). My customers were called "service bureaus." They provided outsourced application services like Medicare claims processing systems to their end customers. Are today's server virtualization and cloud really that different?