Evolution of Personal Computers
When I talked yesterday about thirty years of personal computers, I did not mean IBM compatibles. I meant all PCs (including, for example, the Mac, the TRS-80, and the Amiga.) I did not make that clear, and I apologize.
Here is a time-line of major events in the evolution of PC reliability. Technically speaking, I mean protected memory, preemptive scheduling, and allied protection features which keep an error in one application from crashing other applications or the whole system.
- 1982. P/OS, Digital Equipment Corporation
- DEC miniaturized an existing data center architecture having protection features (RSX-11M), to compete with the IBM PC.
- 1987. OS/2, IBM
- IBM developed this Windows competitor from the ground up, incorporating protection features they knew to be important from their data center experience.
- 1993. Windows NT, Microsoft
- Microsoft started to catch up in 1990, with limited memory protection in Windows 3.0. They only really got it right when they released Windows NT 3.1. NT was a rewrite of Windows by ex DEC developers.
- 2001. Mac OS, Apple
- Apple tried to rewrite their own operating system, but what ended up working well for them was a merger of their original Mac OS with the UNIX operating system.
(It is hard to know if and when to put Linux on the time-line. Linux had protection features from its inception in 1991 because its design was based on UNIX, but it only now beginning to make any inroads into the personal computer mass market.)
The time-line is based on a bit of research and what remains of my memory, so please correct me if you know better.
Advantages Of The Data Center
A family computer is used for business and personal records. On a family PC you will find a whole lot of correspondence in the form of saved e-mail. There will be various kinds of record keeping, from family histories to monthly budgets. Many people are storing photos and music collections on the computer. Any number of personal projects—greeting cards, posters, newsletters.
People put too much faith in PC hard drives, discs, and memory cards to preserve their family records. These devices break, and they get stolen. Migrating away from local applications is advantageous. Data center equipment breaks too, but data centers have parallel secondary systems that take over until the primary systems are restored, with no loss of data and little or no interruption of service. Security against physical theft of the storage devices is much better.
How Risky Is It to Migrate?
On the other hand there are some new things to worry about. What is the risk of unauthorized, undetected use of family records by data center personnel or by the government? Google has a strict privacy policy, but what enforces it, and, for that matter, what keeps Google from changing its mind? The risk is not easy for the mass market to assess. This fact leads them to imagine that it might be a high risk, and avoid it. Similarly, they find it hard to assess the risk that Google will close its facilities unexpectedly. And the very fact that Google provides free service creates a fear that Google has no incentive to provide a reliable service.
There are real risks here, but I think they are often hugely overestimated. The incentives to Google are being looked at the wrong way. Google is like television in an important respect.
You (the user) are not the customer.
Google is not your supplier.
The advertiser is the customer.
You are the product.
Google earned $1.35 billion in the last quarter and has $14.4 billion in cash. There is plenty of incentive to continue attracting users by creating and maintaining a reputation for ethical behavior and reliability. If Google offends its users, they will go elsewhere and Google’s revenue will go with them.
Google doesn’t only run public services. It also runs the same services at private data centers for paying corporate customers. But corporate demand for high availability causes improvements which carry over to the public services.
When a bank goes bankrupt, its operations do not stop. Another bank buys its operations and its customers. My login screen for WaMu now mentions JPMorgan Chase in passing; it’s the same bank under new management. If Google goes bankrupt, its operations and its user base are too valuable to be discarded. Another player in the cloud computing market will buy them, add their name, and go on. The real risk is that at that point I will be offended by a logo reading “Microsoft Google”, and take my files and go elsewhere.
Supposing I am wrong, and one day Google is simply not there. How will I get my family records back? Well, Google actually stores my documents in an open, non-proprietary format. Google also keeps copies of all my documents on my personal computer if I wish. (And, of course, I have opted to do this.) Google offers this option so that I can continue working if my network connection goes down, but I can actually continue working if Google goes down and stays down, until the open source community takes over for them. I think this is really why Google does so much to support open source software. By making themselves non-proprietary, they eliminate risks that I would otherwise be taking by using them.
This is not to say that people should not keep their critical family records in printed form. Nobody should think of keeping their will in Google Docs. And even paper has its problems. We have a printer with archival quality, pigment based ink. The family photos we print with it ought to last many, many years, long after the plastic compact discs gas out enough plasticizer to self-destruct. But it is still true that a fire, a flood, or a plague of insects can destroy them, and I am glad enough to have copies of them stored in digital form at a data center.
Labels: advertising, Amiga, Apple, archiving, cloud computing, data center, DEC, Google, IBM, Macintosh, Microsoft, open format, open source, PC, preemptive scheduling, protected memory, reliability, security, TRS-80
(go to complete article)
