Hacker News Reader

Comment by dredmorbius | Hacker News Reader

Of course, Sun had its own Cloud offering pre-aquisition (SunGrid). Though it didn't seem to be particularly successful / popular at the time (I had a journalist reach out to me to ask if I actually knew of anyone using it at the time of the merger).

Many of the staff and execs left shortly after -- Lew Tucker in particular, February, 2010.

View on HN

At the time I believe it was called "utility computing". I think they were a few years ahead of the curve, and far too dependent on their own hardware platform, for it to be a viable product.

Amazon hit at the right time: virtualization as a concept was being adopted by developers, storage needs were going through the roof, classic storage systems weren't scaling at reasonable costs.

Amazon's other strength is that they can use the massive revenue they have sitting around for short periods of time to build out their computational infrastructure (turning their operational revenue into capital spending). IaaS took off; SaaS somewhat, and PaaS almost not at all (except in a few well defined areas).

I can't really see what oracle woudl bring to the game. It looks to me like they are just making a play to take their existing customers and move them to their data centers.

I'm not discounting your criticisms at all. Yes, data processing services have existed for a long time -- EDS offered this as a business model on mainframes rather famously. And while good at it, they were also highly despised for their contractual and service terms. I worked in the 1990s at a site making use of Everyone Dresses the Same, and there was a tremendous amount of friction, as well a strong internal incentive to get away from the service, however contractual obligations largely prevented this. A cautionary note to those presently looking at cloud services, though the lock-in's much smaller.

Sun faced multiple challenges, largely being a legacy product with legacy offerings in a world where price point was key, their service commanded a premium, and much existing infrastructure wasn't positioned to properly capitalize on it. That's just off the top of my head, though I'd be interested in seeing what other factors accounted for the failure.

Oracle has the kiss of death for far too much in my book. I'd consider any offering it presented with extreme reluctance.

There were two offerings as far as "SunGrid" was concerned - either the public-facing 'retail' iteration powered by N1GE which was launched as network.com ("$1 per CPU per hour!") and also a B2B commercial version which offered a lot more flexibility, i.e run your Linux distribution on Sun's hardware. The latter was built at scale using AMD Opteron-based SunFire V20z's for compute, for example.

The former certainly suffered from platform constraints and a fairly rigid 'stack', but the latter was far more flexible and various teams within Sun worked closely with several potential customers on the proof of concept. Unfortunately this fell apart thanks mainly due to the commercials and some concerns with how the platform was configured within the context of security and ensuring appropriate segregation of data.

Ah, the SunFire. It's hard to compete with the rest of the world when you design a custom internal bus for your x86 product.

I like that the last sentence you wrote could easily describe today; many customers still care very much about the security context and the data segregation.

Reply on news.ycombinator.com