Last week I sat on a panel at a U.S. State Department event that looked ahead at the future of technology. My assignment: Predict what mainstream IT will be like 25 to 30 years from now.
A friend suggested I simply say the two words "flying cars." But I took the assignment seriously and conjured up my predictions the only way I knew how: Start with major problems IT faces today and guess how they will be solved. Coming up with IT's four biggest problems was relatively easy.
Managing infrastructure complexity. Solutions are emerging, but the biggest problem for most IT operations remains keeping the data center infrastructure reliable and available. The main difficulty is diversity: legacy systems, virtualized resources, specialized infrastructure, the rollover from obsolete to new equipment and management software, and so on. Plus, all those temporary fixes that were put in place years ago that no one has time to redo properly.
Fighting a losing security battle. The main thing that determines whether criminal hackers breach your defenses is whether you have anything worth stealing. Everyone knows that perimeter security, although absolutely essential, has no chance of fending off determined bad guys. Today, your best shot is to harden security around your most valuable assets and hope that sends cyber looters somewhere else where the pickings are easier.
Developing the right applications. Ultimately, IT's job is to deliver applications the business needs and wants. (This is what the cliché "IT/business alignment [4]" actually means.) While many organizations have improved their development processes, typically the dev, test, and deploy cycle still yields applications that fall short of what business stakeholders expected when requirements were captured.
Dealing with ballooning data. At InfoWorld, we've devoted a channel to the data explosion [5] in honor of a pronouncement made by IDC in 2008 that data in the enterprise doubles every 18 months. It's not just accommodating those petabytes, much of it now security and system metadata. If you're going to store such gargantuan quantities, wouldn't it be nice to fulfill one of the oldest promises in IT and derive consummate value from it?
When you project out how the IT industry and its customers will solve these problems, it helps to remember that much of the technology we use today was developed over three decades ago, such as the x86 instruction set (1978), the Ethernet protocol stack (1980), and the SCSI standard (1978), to name a few.
When you consider the trends and technologies today that have the best shot at reducing IT's biggest headaches, the future gets slightly less hazy than you might expect. Here's my best shot.
All cloud, all the time. The architecture of cloud computing is all about pooled resources and the advanced tools to manage them. The pioneers of that architecture happen to be today's cloud service providers. But will it matter who owns and operates a data center fabric that's fully automated, converged, and virtualized on a foundation of relentlessly commoditized compute, storage, and network infrastructure? The OpenStack and Open Compute initiatives come closest to suggesting that vision today, and if I live that long I won't be surprised to see them persist three decades from now.
Persistent identity, multiple Internets. I agree with Roger Grimes's [6] argument that persistent user identification is the only real solution to our current security mess. So 28 years from now, this will mean everyone who logs on to the Internet will need to be positively identified, probably through some advanced biometric means. But some people may not have that means or find the end of anonymity repulsive. There will be at least one other Internet -- probably similar to the one we have today. Does that mean the evil forces who want to create a tiered Internet with different quality-of-service levels will have won? Probably. Will there be cyber wars between the Tier 2 and Tier 1 Internets? No doubt.
The converged future of applications. Persistent identification and voice UIs are going to change applications forever. Today it makes sense that you use an email app to send messages and an ERP system to generate financial reports. But with your confirmed identity and with cloud-based application integration, you should simply be able to tell your computer you need the last quarter's sales figures and compose a message to Mom in the next breath. Nonetheless, business applications will still be needed, mainly based on workflows unique to organizations. Authorized people will be able to create them on the fly. Yet I also think there will be a tremendous need for developers to create and improve aggregate applications -- and cloud architecture will make dev, test, and deploy vastly easier.
The big data revolution. Storage technology improvements coupled with cloud scalability will stay ahead of ravenous demand for inexpensive capacity. But what we do with that data represents the future promise of computing. The contemporary cliché "data is the new oil" is apt. Cheap computing power will enable us to derive enormous new value from crunching petabytes of data and metadata using distributed processing frameworks such as Hadoop (another technology that may persist 28 years from now). Creative new forms of data visualization will enable entirely new disciplines. And as IBM's Drew Clark told me in a recent interview [7], the value in the forthcoming explosion of sensor networks will be in the big data analytics that give the data collected by those networks meaning.
Of course, if persistent identity succeeds, this totally integrated, transparent, and hypercrunched future will also mean the complete and utter destruction of privacy [8]. Oh well. Scott McNealy declared privacy dead years ago. But if we must all depend on public cloud service providers that are as inept as today's telecoms, god help us all.
I hope you've enjoyed my impersonation of a futurist. Take your own shot in the comments to this post.
联系客服