Wherefore Moore's Law Part 2: Cultural Change by Classical Electromagnetics


Your Users' Expectations of Your Architecture, by Year

Last time we took a look at where Computing might go in the future. Why? Because we would all like to know when we might expect to stabilize our environments, when Computing might begin to resemble more traditional industries like Construction or Manufacturing. The quick answer is Not in our lifetime.

Some of my colleagues occasionally tell me that such-and-such solution is sufficient. That their new tablet meets all their needs. That no-one will ever require more than 640k of RAM or similar foolishness. The Curmudgeon laughs, ruefully.

I drew the beginning of the end of Computing as being when gaming consoles could reproduce and transmit the activity of our entire forebrain. But why would a CIO care about what a gaming console can do? You've got a business to run, after all. I do wholeheartedly agree, but there's a near-insurmountable cultural problem to overcome if you want to get the gaming console out of your business space.

The whole Computing complex keeps time to Moore's drum. Everything doubles and doubles, changes and improves. We are immersed in the advancement of it, and our personal experiences keep the pace, our expectations rise. NTSC, VGA, 720p, 1080p, 3-D 1080p. Atari, Colecovision, Nintendo, Playstation, Playstation2, Playstation3. 286, 386, 486, 586, 686, x86_64. Computing runs through the lives of all of our citizens, and the complex keeps pushing the envelope to bigger and better.

In the meantime, your bricks-and-mortar business probably isn't changing like that. The number of widgets you manufacture, the board-feet of timber you cut, the number of patients you can listen to aren't doubling every two years. So why would your business computing need to double every two years?

Because we arrive at work with a mental picture of Computing that we bring with us from the outside world. Computers at work should be able to do what computers out there can do. Otherwise the infrastructure must be deficient, obsolete, outdated.

Let me draw another example. MYRA Systems can help with the introduction of Thin Clients to your environment. They have substantial rational advantages over "fat" clients. But they have cultural problem because of an electromagnetic principle.

In my example, we had a client insist that their organization could not adopt Thin Clients that were not capable of showing 720p video cleanly with good audio. How did 720p video get to be a business requirement? That's a cultural issue, and surely derived from the consumer expectation rather than from an analysis of their bricks-and-mortar business processes. 720p video is trivial for a personal computer to render - the environment should be able to do it, or it must be deficient. But there's an electromagnetic problem for Thin Clients.

Bandwidth is harder to transmit as distances rise. This is a golden, eternal, irreconcilable problem. It is fundamental to the way the universe is built, and Moore's Law isn't going to change it. Let's look at that table of data rates again (in Gbps):
 

Data Rate (Gbps) Machine Function
1638 PlayStation3 Cell Processor Element Interconnect Bus
480 PC3-20000 DDR3 RAM
409 HyperTransport 3.1 System Bus
21.6 DisplayPort 1.2
10 Ethernet / iSCSI / FCoE datacentre
1 Ethernet Desktop
0.1 E100 WAN link / DOCSIS 3 Cable Modem
0.05 802.11G WiFi
0.025 ADSL2+
0.013 3.5G Cellular
0.0014 CD Audio

See the relationship between distance and bandwidth enshrined there? On-CPU is about 4 times faster than CPU-to-RAM. That's about 5 times faster than getting to a local video screen. Local video is about 20 times faster than a LAN link. A LAN link is about 40 times faster than a WAN link. As distance rises, bandwidth drops - dramatically.

Thin Clients are predicated on keeping their computing power centrally shared and administered - if it wasn't centralized, there wouldn't be the administrative advantages to be had. But their downside is that the link between the processor in the datacentre and the display device at the desk is slower than between a PC's processor and its display device. And that relationship can't change. When the LAN link speeds up, the processor will have sped up, too. They're both marching to Moore's drum, but the length of their legs are eternally different.

So the trivial-for-a-PC 720p video is a heck of a challenge for a Thin Client. The cultural expectations our client's workforce arrived with butted straight into an electromagnetic hurdle. I'm pleased to say that MYRA Systems overcame that challenge, but it illustrates the difficulties.

For most Computing clients, the right answer for these expectations is to fulfill them - regardless of their applicability to business process. Cultural resistance to anything else is just too difficult, otherwise. The CIO must care about what the Playstation3 is doing -it will shape his or her tomorrow.

That said, it is not impossible to march to a different drum. Take hardened military IT, or ruggedized field terminals in Forestry, or simple Point-of-Sale systems - I saw a vt220 green-screen CRT terminal in Mayfair Mall the other day! Such systems are designed towards a business purpose, and their users come to expect that they will not resemble their home PC system. The lifespans are commonly much longer than generic Computing platforms. And despite their initial costs, the lifecycle costs are generally lower.

If your medium term IT strategy is delivering anything less than the latest equipment you can buy at Future Shop, you're going to have a cultural problem. And if it's not locally-processing and fully cabled, you'll need to add your electromagnetic problem. We see this all the time.  Why else would MYRA have an Organizational Change Management practice?