This week I was searching on the web to see what the technology mega trends are. I want to make sure I have the right tech to sell after-all. Disappointingly no-one seems to have done the thinking for me. Even the "smartest guys in the room" had a disappointing report focusing in on random transient tech.
So I've pulled out my crystal ball - well ruler really as I'm more into extrapolating, to share my strategic view on tech trends.
1. Power
The world seems to have insatiable demand for data. All these blogs, tweets and billions of new web pages need to be stored somewhere. That storage needs power. Cloud means data centres are getting bigger. Google, Rackspace etc are building their data centers in the not so high tech state of Oregon.
Why?
Cheap power, mild climate and tax breaks. Oregon has hydro electric power hence cheap power. Mild climate means less power for air conditioning/cooling.
For years power density of equipment in data centres has been a problem and the percentage of power needed for cooling has steadily risen. Equipment power density continues to rise - the good news is work done for power is improving. Only challenge is we want more things done.
If companies in the USA are being forced to locate data centres in tech deserts simply to exploit availability of power then it is reasonable to assume this trend will only get worse. How long before Google become derivatives traders in oil and energy commodities simply to secure their source of power?
Any tech which reduces energy consumption in data centres has a future.
2. Wireless
Poor old Intel and Microsoft have a problem. Tablet/smart-phone sales (using ARM, MIPS, android or iOS) have overtaken PC sales. There's no old school tech inside.
The challenge to the old giants is not the only thing here. These devices are wireless and they dont have much storage. So for my first trend - wireless.
You might have noticed but generally smart phones and most tablets dont have an Ethernet jack and in my case I havent got an Ethernet jack in the bathroom anyway and I also use mine on the train too - the cable isnt long enough either. I therefore tend to rely on WiFi or 3G (soon to be LTE).
So as more people buy these devices there are more-and-more things on wireless. I dont live in a particularly densely populated area but the number of WiFi APs near my house has quadrupled in 2 years. Given most of my neighbours are old this means there are lots of silver surfers....
All these WiFi APs are now making my WiFi less reliable since it is getting cluttered. Given I have about 20 devices on my AP (ie a legacy) it is not feasible for me to easily migrate to the less popular 3.5GHz WiFi to escape the noise.
If the trend continues, WiFi is going to collapse under it's own popularity.
Any tech which makes WiFi more reliable has a future.
Whilst I'm on the subject of WiFi, it really needs some structure. I find it great as a home technology but it needs some serious rethinking as a public service. It needs proper roaming and why would I pay BT Openzone £5.99 for 90 minutes when I can use 3G which costs me £5 per month ...
So that brings me to mobile. 3G mobile Internet is great and LTE looks like it will be even better (except probably battery life). However the operators have a problem. If people are ditching their computers then that means they are doing more on their phones/tablets. Great for sales in the short term but it does mean traffic demand on the mobile networks will continue to rise. There is already a major shift in traffic type from web browsing to watching data hungry videos on phones which is creating a step change in data volumes. This means investment. However more traffic doesnt equal more revenue or more profit for the operators.
Besides the problem that wireless has a finite capacity and all these devices will consume that capacity, the cost per bit transported is not falling fast enough. Something has to give. Unfortunately the likes of Google and Apple have done such a great job of eroding mobile operator profits from value added services that ironically they are choking to death the very companies that they depend upon.
Prices need to rise (unhappy consumers), suppliers need to come up with ultra low price tech for lower cost per bit or the operators need to figure out how to exist on very low margins.
I predict that this will reach crisis point and bandwidth eco-system will collapse (probably taking modern society with it for a while). This will force the introduction of a flow of cash with the flow of data - something which old world telephone calls had but the internet hasnt got. Companies like Facebook, Twitter, BBC, NetFlix will start to have to pay for the bandwidth end-to-end their customers are consuming.
3. Storage
Demand for data storage continues to grow. Increasingly we are storing our data on the cloud (whether we know that since the apps on our tablets do it automatically anyway). Hard-disks still get bigger. 3 tera bytes drives are becoming the norm. However hard drive technology is starting to reach the limits. The trusty magnet has taken us a long way but the problem is size of the magnetic particles is reaching some physical limits. The most recent solution has been to flip the magnets orthogonally so the cobalt magnets on the disk platter are vertical allowing the next step change in storage. There isnt an obvious next step. Increasing magnetic hard disk sizes will come to a halt in the next few years. Solid state drives are orders of magnitude smaller than magnetic hard-drives and probably wont catch up in the same time frame that magnetic hard drives reach their limit.
Unless there is rationing of tweets something has to give. The solution is more hard-drives. Data storage density will reach a plateau and more hard drives will be needed - the cost per byte wont fall. Data centres will need more power for these hard-drives - they wont be able to rely on the falling watt per byte trend.
If anyone knows of any great new storage technology (such as molecular or genetic storage arrays) let me know - it sounds exciting new tech to invest in.
4. Processors
Today's processors have a problem - the real world is just too slow. The CPU cores today are usually far faster than the peripherals or indeed the I/O on the chip. Why is this a problem? Well it does mean that computers need to become more parallel. We have already seen this trend with multi-core processors. Today's super computers are in-fact massive distributed systems. It is getting harder to build computers that "scale-up". It is often far easier to "scale-out" than "scale-up"- use lots of relatively inexpensive computers in parallel. Google adopts this approach for example.
Developing systems which are massively parallel is an important trend. Processors will get more cores but it does mean a paradigm shift in programming. Most programming languages are not well suited to exploiting multiple-cores or multiple machines.
Programming languages like Erlang, Scala will therefore rise in popularity as a result.
Enough typing for today. Next week I'll cover Payments, Networking/Bandwidth including home networking, Internet of Things and security.
No comments:
Post a Comment