We're hearing rumours that Oracle is ending the Sun-LSI relationship whereby Sun incorporated drive arrays from LSI. When Oracle bought Sun it inherited two storage OEM deals. One was with HDS for 9900 high-end arrays, based on its USP-V product, and has been terminated. The other deal saw Sun's mid-range 6000 storage products …
Good Riddance to LSI
You would have thought Sun would have learnt their lesson with their SE35100 arrays which they OEM'd from Dot Hill where you got what you paid for - cheap, flaky, unreliable and rubbish support from Sun. Then they started selling the SE6140 line OEM'd from LSI that were similarly cheap, flaky, unreliable and came with rubbish Sun support. The number of arguments with I've had about both the 3510 units and now the 6140s not being fit for purpose doesn't bare thinking about.
LSI and Oracle
A bit cranky this morning are we? No product is perfect but unless you expect each company to completely redevelope every bit of technology OEMing is the only really viable way of providing complete product lines. Sun has developed more of its own technology than many other companies but that obviously hasn't worked out all that well for them since they no longer exist and are now part of Oracle.
Without any doubt the SE35100 had issues but the SE6140 is actually light years better than that product and is reasonably free of issues though the last major code update for the product did throw in quite a few issues that took some time to sort out.
Sun support is no better and no worse than support from other large companies like this and is more a function of who you end up getting to work on your problem. But, considering the years and years of layoffs that went on in Sun it is amazing anyone even answers the phone.
But even if Oracle wanted to get rid of the relationship if you look at the last published fiscals statements for Sun you see that the revenue of the LSI product line is about 10 times that of their own internal products and I don't see Oracle cutting off that revenue stream any time soon. But I would never make a bet one way or the other on what Ellison will or will not do.
Different experience than "Good riddance..."
Not surprised if they cut off all OEM agreements as they stated with HDS and Netbackup. However our experience has been fairly good with the LSI arrays through the generations.
We've several 6140's, 6540's and 6580's and they have stood up fine compared to outages on Netapp and EVA's that we've had. Performance has flown over the Netapp as well despite the price difference (which I find comical, it's a very real price difference) & yes, I'm comparing apples to apples (SATA to SATA, Fibre to Fibre, 15k - 15k etc)
On this experience I'd say it's a shame, the sales IOP's claims seem to be backed by real life experience & have been reliable.
Agree with the 3510 dothill series. Early firmware problems rendered them unreliable to point of shame on Sun for selling them.
Always could be luck though, your looking at a sample population of 1 here.
Not good riddance
The 3510s did have their set of issues with older firmware. Never had a problem with any of the LSI rebranded arrays. StorageTek(Sun) claimed they added extra engineering into their rebrand. Not sure how true this is. What you also seem to forget, is that it's not just Sun that OEM'd LSI arrays. IBM also does for a hefty price.
I can see them dropping LSI like they did with HDS since the unified storage is a good product line and makes them more cash. If Diskcrash is right in his findings on Sun/Oracle making money off the OEM deal, they won't sever it anytime soon. There's also still a lot of people that prefer the traditional storage model, so they should keep the OEM deal for the entry to mid-level storage alive.
I know that line is uses LSI controllers, but I'm not sure if they need to keep the OEM deal alive for the JBODs they sell for the unified storage controllers. Anyone know if they're solely designed by Sun, or if it's part of an OEM deal?
You can still get them
From IBM, same thing, different badge, cheaper and with the Windows VSS support, etc that Sun didn't OEM...
What will take it's place?
I am wondering... what will take LSI equipment's place?
re: What will take it's place?
I would say nothing will take it's place directly.
The main general purpose storage offering will now be the S7000 range, now that it can be fibre channel AN attached since the last software update.
Larry hates OEM deals and Fujitsu
This is why Oracle is not even talking to Fujitsu about APL2. SPARC64 VIII is canceled.
LSI still exists?
I thought that company disappeared eons ago. I guess they are just largely an irrelevant shell of their former selves ala SGI, Atmel, Freescale, etc.
Became Engenio (possibly after they swallowed Symbios), then changed the name back to LSI Logic later. Probably because it's a better name.
LSI still exists
You might be thinking of a different company (Adaptec maybe?) they are not the same company that they were a number of years ago and have focused on storage as their niche. In addition to OEMing storage for a number of manufacturers they also have the Mylex, Megaraid and 3Ware raid controllers and own a large percentage of the SAS and raid on the motherboard market. Here on TheRegister there was just a PR blurb from them about doing 1million IOPs on PCI-SSDs.
As to what Oracle would replace the LSI kit with well that's easy since if the reason to drop it is that you don't want to OEM third party kit will mean that you plan on using internally developed products which in this case would be the 7000 series. That would make a lot of sense in that Oracle would have higher integration and tighter control of the final product. It wouldn't make sense in that at the moment the 7000 has some performance/cost parameters that makes it hard to cover the entire entry level to enterprise level range that they would need if they really plan on being competitive.
If they wanted to replace it with some other third party there aren't a lot of good options for them there.
Realistically Oracle can be expected to maintain their relationship with LSI for at least the next few years no matter what they might want or being planning for the future.
Of course LSI still exists
Of course - most of the Oracle / Sun and IBM storage sales is based on LSI.
This Reg article is nothing but a Troll / trawl by the way.
>SPARC64 VIII is canceled.
Good! It was a long overdue mercy killing. Not a huge fan of developing on Solaris but at least on x86_64 it is almost speedy enough to be tolerable. I absolutely loath having to work on the SPARC boxes at work. Such dogs in cpu and io performance to the point of feeling like I am working inside a linksys embedded router or something. Die SPARC die!
Now don't get your panties in a twists there young pc programmer.
That was just a extrapolation of an extrapolation of nothing that lead to some wise ass thinking Sparc was cancelled.
Fuj still has a large investement in Sparc and teh killing of the Rock has helped them a lot.
SPARC64 VIII is canceled
canceled gone...dead....history....just check Fujitsu's roadmap...or oracle's fujistu roadmap..end of story...
oh thats right....rock has never been officially canceled either
Actually not so young
I even got to play around with Cobol and JCL back in the 90's. As for SPARC yes in the 90's it was much better compared to its peers and the open source architecture of the chip was a great idea (though a panic move on Suns part), but basically big business can not run away from SPARC fast enough right now. In fact much of the work I am doing is moving code from SPARC to the much cheaper and faster commodity x86_64. With Intel adding RAS to the Xeons the mid to big iron vendors are really going to start having to justify the insane margins they have charged in the past (I am looking at you too IBM and your POWER architecture. Better keeping winning game console contracts or else chipzilla will bury you too).
I believe this puts the speculation to rest...