Sun Microsystems has firmed up Fujitsu as a Solaris reseller on x86 boxes, which leaves us wondering when HP will break and join the rest of the crowd. Sun this week announced the deal with Fujitsu Siemens, which will see Solaris offered as an option on some Xeon-based Primergy servers. Fujitsu joins other Tier 1 server vendors …
Surprised that Fujitsu took so long
considering that they had an existing relationship with Sun.
HP is holding out because it is likely to further decrease the relevance of anyone doing anything on HP_UX/Itanic.
Is there a fujitsu unix?
There is HP_UX, which may explain HP not being too bothered about jumping to Solaris x86.
Solaris use to be nice on Sun equipment, and sure x86 with multiple redundancy machines is the better way for a tech company to operate now, Sun has still got be looking to make money on selling their hardware kit. Which tends to fit into an organisation that only has a few good tech people.
It sort of pops up on your radar when you have to admin on or develop for the Solaris platform. And then really who cares who you get the x86 equipment from, sure HP will get some business but I am sure a slew of other vendors will as well.
Sun do x86 platform at the low end, but you get Solaris bundled, and the deals are ok but nothing great, there is no overriding reason to get a solaris x86 machine, that is perhaps where they need to do some work on. Make a niche in the market for themselves their as well.
Solaris on x86 is a non-revenue activity
Except Fujitsu finally supported it to get something else from Sun. Sun likes to trade Java licensing for Slolaris on x86 support
Getting it by the back door.
Don't forget that HP's slated (court cases permitting) to take over EDS - which is a big Sun shop. So maybe this'd give HP an excuse to sell their soul to the Solaris devil and not lose too much face?
But then again, anything would be better than HP-UX, maybe this is why HP push Linux these days? And no, I'm not a Solaris admin, nor Sun employee. :)
(Just donning my flame-proof gear waiting for the slagging from the HPUX fanboi's... <grin>)
Why use Solaris x86
a) You have an in-house Solaris application and can recompile it to go onto x86 - not many people in this position
b) You have a number of Solaris admins and want a x86 *nix. Saves cross training them to a GNU/Linux distro, though there is still some cross training.
c) Your datacentre is Solaris/SPARC, and you only want a little x86 *nix
d) You need to have really effective SMP on your x86.
e) You have got fed up with the Red Hat upgrade treadmill and want a x86 *nix with longevity (this is becoming the No 1 reason)
The reasons why you might not use Solaris x86 are:
a) You have admins skilled in GNU/Linux
b) You need to use a proprietary application which isn't available for Solaris x86 (e.g. Oracle 11g ATM)
c) The commercial GNU/Linux distributions have a much bigger installed base, so you want to "go with the crowd"
d) You are worried about hardware compatbility
e) You are worried that Sun might go bust and leave you high and dry.
This announcement is about removing inhibitor (d) for another set of machines.
There never was a Fujitsu unix, but through its (now gone) subsidiary HAL fujitsu built vast UltrasparC-based machines (as many as 2048 processor models) running Solaris.
Kinda proved the Scalable part of the Scalable Processor ARChitecture.
Re: Is there a fujitsu unix?
Yes - it's exactly the same as Sun Solaris apart from a tiny handful of extra commands (such as the incredibly useful - to me at least - /serialid). The only other difference is that you are supposed to get your patches from Fujitsu so when an emergency fix comes out, you have to wait a short while for Fujitsu to get the Sun patch and then test it before they release it themselves.
Re: Thomas Kenyon
I seem to recall Siemens having their own Unix at one time, which I assume Fujitsu acquired as part of the merger. It's name was some sort of play on Siemens and Unix - something like SINIX.
RE: Re: Is there a fujitsu unix?
FSC's whole thrust was around making systems that ran standard SPARC Solaris binaries faster for less cash, using their SPARC64 chip, and it worked very well compared to the weak UltraSPANKED kit, which is why Sun have filled the hole between UltraSPANKED and "Rock" with badged FSC SPARC64 servers. This is all the more amusing considering how hard Sun used to FUD the FSC servers. Even then, I have yet to come across a single instance where Solaris on anything outperformed the equivalent hp-ux on Itanium, and I have done POCs on a lot of software stacks to prove that to management.
Three things about HP and Solaris x86:
1/ It's supposedly an open OS, so there is no real reason for HP to have to get a Sun support agreement. In fact, anyone that wants to can set themselves up as a Solaris x86 support business, just as anyone can set themselves up as a generalist Linux support company. And, as HP has both a wealth of experience supporting Solaris (HP has taken over many outsourcing deals where they have taken over administration of existing SPARC kit - I should know, I've arranged one!) and migrating off Solaris onto Itanium, and years of experience with real Linux which Sun can't match due to their on-off love-hate relationship with Linux, HP has more than enough knowledge to support anyone silly enough to want to run Solaris x86 on ProLiant. Which brings us to the second point....
2/ Solaris x86 is a complete non-entity in the OS market. Linux installs are massively more popular, and HP's Linux support revenue dwarfs all the other vendor's Solaris x86 support revenues by a mile. Customers simply don't want Solaris x86, they prefer better products like RHEL or SLES. Maybe HP will show some interest if Solaris x86 ever manages to get out of the lab toy stage.
3/ And lastly, the best point of all - HP can already give you the fastest Solaris system in the world, by using Transitive's emulation software to run standard SPARC binaries on top of Red Hat on Itanium or x86. Given that this is a much easier migration than the kludge of trying to get SPARC apps onto incompatible Solaris x86, and that it puts you on the market-leading kit, please explain why anyone would show an iota of interest in Solaris x86 for real business?
At best, Solaris x86 is a training tool. Sun's best hope would be to ditch "Rock" before it gets waaaaaaaay further behind schedule, dust off and update their old port of Solaris to Itanium, and start shipping it on badged FSC Primequest 500 range (Itanium) servers. Then they might actually stand a chance of being around long enough to develop Solaris x86 into a real competitor to RHEL or SLES. I wouldn't hold my breath though....
Good reason to use...
Solaris x86 is ZFS. Surprised no one mentioned it.
Re: Matt Bryant
Let me get this straight:
Itanium > Rock...
I agree with your second point, though, and u sound like a reasonably informed poster, so you probably remember the history of the re-emergence of Solaris on x86 circa 2002...
it seemed to me when the panic button was hit, at Sun, about this timeframe, to try to stem the tide of Linux, there might have been different options other than saving Solaris as a volume server play...
a. invest in Linux
b. take a position with Red Hat
c. have Oracle do one of the above
d. Focus on middleware revenues on SPARC
a) would have been a good move; b) would have been a temporary move; c) would not have worked; d) would not have stabilized the revenue situation, so i understand that in those desperate moments, "saving" Solaris seemed like the best bad option,
but now we have openSolaris and continuous propaganda around its supposed relevance, and i just don't see why Sun thinks an OS is so mission-critical for a competitive advantage? do you?
RE: douglas dooley
"....i just don't see why Sun thinks an OS is so mission-critical for a competitive advantage? do you?...." It's all about support and services revenue - if you can persuade customers that your proprietary OS and hardware has some business advantage they will buy it, and if you can then convince them only you have the skills and knowledge to support them then you can charge them a steep rate. This is what fuelled commercial UNIX and the mainframe bizz. Linux changed all that - suddenly, the vendors had to try a lot harder to differentiate. Sun has left it too late and doesn't have the ability to differentiate on x86 with Linux or Windows. Their SPARC bizz is dead on its feet, having been out-innovated in the enterprise UNIX market by Itanium and Power. And don't mention their farcial entry into the storage market! So Sun are desperate to try and create an aura of mystical capability around Solaris x86 as a last chance of securing some form of proprietary and therefore controllable revenue stream.
u can fool some of the people some of the time but not all the people all the time
I guess this is why Sun stock is at a 5 year low and IBM is at a 5 year high
Basically, Matt, x86 is a cloud maneuver, but it doesn't reflect the enterprise market, nearly as much...i honestly don't know what u r talking about with Itanium and Power, those r the 2 least likely platforms to make a difference, IMO...
Solaris as a services contract revenue stream seams far-fetched, as well, as they have licensed it to their biggest competitors, leaving a saturated market, in theory, for developer up-take, but if analysts really did their job and gave us the penetration of openSolaris to Linux, we would fall off our chairs laughing at the disparity...
I really think whether it is fiscal year Q1 09 or Q4 10 does not really matter, Sun will have to ship their most valuable and highest priority servers on Linux, and most likely Red Hat, unless Debian pulls something out of nowhere...
it is going to be a battle for solutions on top of Linux with JBoss and Glassfish battling for relevancy, and ultimately splitting the Enterprise Java market where Sun makes most of its revenue; i seriously think that Jonathan's job is on the line the longer he holds on to this Solaris pipe-dream...
does anyone out there know the impediment to Sun getting a Red Hat Enterprise Linux contract on the GSO price list? I know it would be an uphill battle for Glassfish to fight JBoss on Red Hat's terms, but I can't think of a single Unix customer that is not going to be a Linux customer within 5 years...
its just done, fighting the Solaris-battle is the definition of a losing war...