> Smart companies try to commoditize their products’ complements. [...]
> Headline: Sun Develops Java; New “Bytecode” System Means Write Once, Run Anywhere. [...]
> Sun’s enthusiasm for WORA [write once run anywhere] is, um, strange, because Sun is a hardware company. Making hardware a commodity is the last thing they want to do.
Here is a prior conversation on HN:
I think Sun tried to pivot but they couldn't put together a coherent strategy together. Some of their acquisitions were all over the place. Scott McNealy probably stuck around longer than he should have.
Contrast with Apple who had a fairly clear strategy of doubling down on the Mac and then building synergy with other products (like the iPod) to grow.
It wasn't just the obvious price advantage, but despite Solaris' best efforts, Linux was still superior, because people could learn it at home.
First big deal was the megapixel workstation, meeting a science and engineering need. Motorola chips gave way to RISC, (SPARC) which worked well.
Next big deal was the dot-com boom, and timing was perfect for expensive, powerful and reliable computers. Larger systems and networks of Suns powered many dot-coms, and even Pixar. [0]
100 E-4000 servers.
But dot com faded (crashed) and compute farms were built on cheap redundant hardware.
I was systems engineer in education sales, and it got tough. After I left, Sun eventually sold to Oracle.
Sun bought Cobalt for the Qube, which was pretty cool for education but schools below college level ran on “educational” windoze apps, like the infamous “Reader Rabbit”, so we ended up with windoze servers running them and accessing them via Citrix to Sunrays, one per teacher. I’d have much preferred Linux apps ported to Solaris, the likes of the very amazing GCompris set of educational apps.
There’s a TLDR kind of summary at Slashgear. [1]
Sun pioneered lots of great technology, or advanced it. Some include Java, ZFS, and virtualization (Solaris Zones and E-10000 ) and more.
I supported university computing before joining Sun, and it was a good fit. Sun had donated a raft of workstations. And universities made good use of the gear, such as NOW [2], the network of workstations. Arguably the progenitor of Google and Facebook style compute farms.
[0]: https://www.hpcwire.com/1998/12/18/pixar-used-sun-systems-po...
[1]: https://www.slashgear.com/1352662/forgotten-computer-compani...
1) Sun made expensive gear. Nice stuff to be sure, but too expensive for home use. Target market was therefore "business".
2) Intel improved quickly. So every couple years Intel machines basically doubled in performance. When your hook is "fastest" you need to iterate very fast to stay ahead.
3) every time Intel got faster, some part of the market got lost. Which means convincing existing customers their hardware is obsolete and time to buy again. Folk are being sold really expensive gear and then told its obsolete barely a few years later.
4) dot com bust floods the market with cheap everything, including cheap Sun hardware.
5) network effects (software availability, using the same computer at work, at home, at school, experience at prior wotkplaces) all work against Sun. (In another thread I see an ex-Sun person referring to windoze- if all else fails, try shaming the client into making "better choices".)
Java (write once, run anywhere) was a play to improve software availability, but folks selling Java software weren't really interested in supporting niche platforms (so yeah, it probably runs on Sun, but we don't have one so we don't support that). Plus Java became "write once, debug everywhere".
In short Sun lost because Wintel was everywhere, and Sun was too expensive, and ultimately had no edge. In a market with network effects there are winners and everyone else fails.