4 ms·
The tricky thing about that is that most highly-reliable distributed systems these days evolve from startups that are not-at-all-reliable. Tandem's technology
by nostrademons 10y ago
The tricky thing about that is that most highly-reliable distributed systems these days evolve from startups that are not-at-all-reliable.
Tandem's technology is incredible, even by today's standards and certainly by the standards of 1974. But they were designed for a world where you sold computers (hardware + software) to big established companies for millions of dollars. Once you saturate that market (which Tandem, Stratus, DEC, IBM, etc. did by the late 80s), what's next?
Since the mid-90s, most tech innovation has been driven by creating new companies, new networks, and new value chains to replace the old incumbents. This process is inherently experimental, and requires that you start out with a small system that grows into a large system. Buying a Tandem computer for the sharing-economy company you're starting out of your apartment is a non-starter; even if you could afford it, it couples you too tightly to a set of applications that probably isn't relevant for what you want to do.
That's why most of the interest in distributed systems today is in taking clutzy single-box programs that used to be Rails or Node apps and then turning them into robust, fault-tolerant systems that serve billions of users and never go down. That's where the money is. The market of existing big companies that will pay a lot for software is largely tapped-out; the upcoming growth is in working for small companies that are rapidly becoming big by replacing those incumbents.
- Ericson2314 10y agoBy your own evidence, s/most tech innovation/most business success/
- nostrademons 10y agoWell, it's innovation along a different dimension. There's been a massive amount of progress in terms of bringing computing to the consumer since 1995 - just compare the user interface of an ATM to that of a mobile app. There's been a concurrent slide backward in attributes like performance, reliability, and especially security. The market seems to reward ubiquity more than any other dimension right now, and you get products to match. Much of the complexity in often-derided web & mobile app development comes from supporting features - like the back button, direct linking, social interactions, real-time response, and no-install usage - that directly benefit ubiquity, but come with penalties on performance/reliability/security/etc.
- marssaxman 10y agoBringing computing to the consumer in a way that lets you control their experience and extract rents, perhaps. We haven't been doing so well when it comes to the original liberatory purpose of personal computing.
- nickpsecurity 10y ago"Bringing computing to the consumer in a way that lets you control their experience and extract rents, perhaps." Desktops are more open and capable than ever. A number of those capabilities started in rent-seeker platforms, too. So, parent's point about increase in capabilities for consumers stands both for proprietary and FOSS.