ORACLE: New Roadmap for SPARC and Solaris

in hardware on 2014-04-07 17:33 (#3HK)

Oracle has published a roadmap [PDF] for its SPARC and Solaris platforms and makes a commitment to SPARC until at least 2019. This was picked up by The Register in a recent article :

Optimists will be pleased that Oracle appears to have three generations of silicon in mind and appears committed to SPARC until 2019 and beyond. Pessimists might wonder if the "core enhancements" due around 2019 represent a worryingly vague statement of direction and/or intent. The planned 2017 and 2018 enhancements to thread strength and throughput also look incremental compared to the heftier enhancements planned for 2015 and 2016, which could be worrying for those who need SPARC to scale.

[Cross-posted at SoylentNews.org]

They should worry (Score: 2, Interesting)

by fishybell@pipedot.org on 2014-04-07 17:49 (#10E)

If you're truly invested in SPARC, and you haven't already considered jumping ship to x86, I'd recommend reconsidering. Yes, it will continue to bump along, but since Oracle's main focus is their database, which runs fine on x86/Linux, that is where the future is for them. I am not aware of anyone using Solaris for anything other than Oracle and SPARC for anything other than Solaris.

Re: They should worry (Score: 2, Interesting)

by Anonymous Coward on 2014-04-07 18:15 (#10F)

SPARC and their entire stack from ILOM to Solaris is orders of magnitude more robust than anything else out there. They also proved that they can deliver reliably in contrast to IBM where any Power7+ and Power8 are all delayed lately. I think we can expect stories again from people migrating /to/ the Oracle platform in the very near future.

Re: They should worry (Score: 1)

by dnied@pipedot.org on 2014-04-08 07:08 (#10J)

I hope you're right, in the interest of platform diversity.

Re: They should worry (Score: 1, Insightful)

by Anonymous Coward on 2014-04-09 14:48 (#116)

x86 is not so bad and personally I don't appreciate platform diversity with ARM. I would prefer however Power or SPARC (sparc even as an completely open platform). Just compare the source trees in the Linux kernel of sparc and ARM, the latter is a total mess.

Re: They should worry (Score: 2, Interesting)

by billshooterofbul@pipedot.org on 2014-04-08 16:50 (#10N)

Back in the first internet bubble, Sun gear was mandatory for every startup. Then startups migrated to x86 for the next wave, and now they're almost all on AWS. With Sun/sparc not being purchased in such quantity anymore, they are ever more reliant on existing companies replacing older systems and.or adding capacity. I'd love to see Sparc and Power survive and thrive, but it doesn't look likely.

Re: They should worry (Score: 2, Interesting)

by Anonymous Coward on 2014-04-08 17:09 (#10Q)

Oracle customers are not startups, not by a long shot. While I agree with you that Power is the new Itanic and basically dead (several indications; no roadmap, latest chips were late, they created something called 'openpower', ...), for Sparc it might not be true. Fujitsu is also investing and Japans supercomputers use those Sparcs and they probably want to export these as well. Oracle has a huge customer base and as soon as their database will run much faster on the Sparc gear, those customers won't care what ISA they're running anymore.

Re: They should worry (Score: 1)

by billshooterofbul@pipedot.org on 2014-04-08 22:57 (#10V)

Yeah, I just don't know if there is enough revenue in being Oracle's preferred hardware choice to keep sparc a float.

At the end of the day the old formula must still hold: profitMargins x numberOfSales >= costOfSupportingDevelopingSparc.

Re: They should worry (Score: 1, Interesting)

by Anonymous Coward on 2014-04-10 11:30 (#11G)

Whenever that formula doesn't hold, it's called investment ;)