My guess is that they'll put Solaris in maintenance mode until Linux becomes accepted as a high-availability, enterprise platform. At that point, there would be no reason to maintain Solaris; companies would move to Linux anyway.
It costs money to keep modifying an OS with competitive improvements. Oracle is a database/services company. What do they gain by spending money on their own proprietary OS?
Oracle is not abandoning Linux because it may be the enterprise OS of the future. I have no idea what they plan to do with their RedHat agreements. Oracle short-term will keep Solaris to keep Sun's enterprise clients.
Most of Sun's revenue comes from their Sparc hardware, even though sales have been declining.
They need Solaris for their Sparc servers and since the x86 and Sparc versions come from the same codebase, and the x86 server sales are increasing, it doesn't make sense to ditch Solaris.
One of the reasons Sun's became such a dominant player in the unix market (especially considering their relatively small size) is that, in addition to buying Sun's hardware on the merits of the hardware, a lot of people would buy sun hardware to be able to run Solaris. The same is not true for HPUX and AIX. While there are some fans of those OSs, they dwarf in comparison to Solaris.
Oracle wants sun's hardware business, including SPARC. That means Solaris isn't going anywhere.
They need Solaris for their Sparc servers and since the x86 and Sparc versions come from the same codebase, and the x86 server sales are increasing, it doesn't make sense to ditch Solaris.
Two little details: First, there's a SPARC version of Linux, though Sun doesn't support it. (Cannonical does.) Second, most of Sun's x86 servers end up running Linux or Windows. A lot of Sun people are in denial about that, and still pretend that the x86 systems are still just a migration path to SPARC; Oracle will have no reason to continue that pretense.
That said, I do agree that Oracle is likely to keep Solaris going. Oracle database does run on Linux, but Oracle people tend to prefer Solaris. It's proba
Two little details: First, there's a SPARC version of Linux, though Sun doesn't support it. (Cannonical does.)
I had Debian running on an old SPARC box a few years ago, but Linux on SPARC is not a viable alternative for enterprise deployments. While Cannonical may support it, it's not going to be supported by most ISV's. I don't even believe Oracle has support for Ubuntu. In the enterprise data center, if you're talking linux, you're generally talking RedHat. That's where the ISV support is. RedHat no longer supports SPARC.
Second, most of Sun's x86 servers end up running Linux or Windows.
I've said this in other threads. Oracle gets deployed on Solaris/SPARC more than on Linux acco
As a matter of fact, I mostly agree with you. I just don't agree that there's any connection between Solaris's survival and the survival of Sun's x86 servers.
But yeah, the Oracle takeover is the best thing that could possibly happen to Sun's x86 products. Until now, they've been limited by the SPARC/Solaris Everwhere! mindset that still dominates Sun. (Better than it used to be — remember Cobalt? — but still pretty bad.) Now these servers will fall into the hands of a huge sales force that has n
Last reported quarter, Sun had around $916 million dollars in UltraSPARC/CoolThreads related hardware billings and only $176 million in x86 servers. The SPARC servers are supposedly also higher margin than the x64 servers.
SPARC sales are declining, while X86 servers (and CoolThreads) are increasing, but that's still a lot of money for SPARC. Add to that whatever the share of the $900 million in total hardware/os support fees that go towards SPARC based servers.
I mostly agree with your analysis (the SUV analogy is particularly apt) but I think you're making a very flawed assumption: that Sun's share of the x86 server market is as big as its going to get. As I've already argued, Oracle is in an excellent position to grow that market drastically.
very flawed assumption: that Sun's share of the x86 server market is as big as its going to get.
I'm not making that assumption. I think the x86 market will grow for Sun and I'm really surprised they haven't seen greater growth in that line. They have good designs and on their lower end, are price competitive with Dell. They even offer some larger servers and their relationship with Intel and AMD help them get them out to market fast. They were the first to come out with an 8 way quad opteron server. The x4600 [youtube.com]. And look. He picked up the server by the back of the chassis and didn't start bleeding:) O
To add. I also thing Sun has been getting a very raw deal.
Everyone is painting IBM as such the Linux supporter, but when you get IBM salespeople in, AIX and DB2 come up in the conversation.
IBM and HP have had Solaris to Linux migration campaigns, but not AIX or HPUX to Linux migration drives and if you search for AIX vs IBM you'll find some intersting stuff on IBM's site.
Solaris was (maybe still is) the leading Unix. HP and IBM didn't say Linux was better than Unix. They said Linux was better than Solaris a
Ah here is the sensible reply I was looking for. Now onto speculation:
As an example Oracle9i runs fine on a block device instead of a fs. The OS only seems to get in the way in the eyes of Oracle DB. That lends credence to the idea that Solaris vs. Linux is not worth the extra dev and maintenance costs. After all Oracle Corp already can sell you a turn key solution involving x86 boxes, Linux, and 10g. One thing that people running Oracle DBs on Linux wish they had was truss and dtrace though. Oracle Corp ma
> As an example Oracle9i runs fine on a block device instead of a fs. The OS only seems to get in the way in the eyes of Oracle DB.
Not entirely true.
1. Running Oracle on a block device instead of a filesystem makes certain classes of backups (like coldsnaps) a real pain in the ass 2. Ever since libaio, the filesystem layer is no longer significant in terms of Oracle performance (at least on Solaris, don't know about others).
ZFS with automatic hotsparing blah blah may actually improve things for filesystem
My brain boggles that you got AIO, VxFS, and Oracle DB to play nicely, wow you don't know how lucky you were or you did it under a certain mix of versions that was relatively stable.
I agree in point 1, but the performance increase was typically 30% or so. Thus that pain had to be dealt with.
The ZFS point, again why I see Oracle Corp seeing a place where there is money to be made.
Or, use a zvol instead of a full-fledged zfs filesystem. You get all the benefits of the storage pool and easy snapshotting, as well as all the benefits of a "bare blockdevice".
Wow (Score:5, Interesting)
This is a big surprise.
Wonder if Solaris will become their main development platform again.
Re: (Score:3, Insightful)
My guess is that they'll put Solaris in maintenance mode until Linux becomes accepted as a high-availability, enterprise platform. At that point, there would be no reason to maintain Solaris; companies would move to Linux anyway.
Re: (Score:5, Insightful)
Actually. I think it might well go the other way. That Oracle decided to fork/clone Red Hat shows one thing - Oracle WANTS to have an OS.
Now they have one.
Re:Wow (Score:2)
You're not looking at the environment properly.
It costs money to keep modifying an OS with competitive improvements. Oracle is a database/services company. What do they gain by spending money on their own proprietary OS?
Oracle is not abandoning Linux because it may be the enterprise OS of the future. I have no idea what they plan to do with their RedHat agreements. Oracle short-term will keep Solaris to keep Sun's enterprise clients.
Re:Wow (Score:5, Informative)
Most of Sun's revenue comes from their Sparc hardware, even though sales have been declining.
They need Solaris for their Sparc servers and since the x86 and Sparc versions come from the same codebase, and the x86 server sales are increasing, it doesn't make sense to ditch Solaris.
One of the reasons Sun's became such a dominant player in the unix market (especially considering their relatively small size) is that, in addition to buying Sun's hardware on the merits of the hardware, a lot of people would buy sun hardware to be able to run Solaris. The same is not true for HPUX and AIX. While there are some fans of those OSs, they dwarf in comparison to Solaris.
Oracle wants sun's hardware business, including SPARC. That means Solaris isn't going anywhere.
Re: (Score:2)
They need Solaris for their Sparc servers and since the x86 and Sparc versions come from the same codebase, and the x86 server sales are increasing, it doesn't make sense to ditch Solaris.
Two little details: First, there's a SPARC version of Linux, though Sun doesn't support it. (Cannonical does.) Second, most of Sun's x86 servers end up running Linux or Windows. A lot of Sun people are in denial about that, and still pretend that the x86 systems are still just a migration path to SPARC; Oracle will have no reason to continue that pretense.
That said, I do agree that Oracle is likely to keep Solaris going. Oracle database does run on Linux, but Oracle people tend to prefer Solaris. It's proba
Re: (Score:2)
Two little details: First, there's a SPARC version of Linux, though Sun doesn't support it. (Cannonical does.)
I had Debian running on an old SPARC box a few years ago, but Linux on SPARC is not a viable alternative for enterprise deployments. While Cannonical may support it, it's not going to be supported by most ISV's. I don't even believe Oracle has support for Ubuntu. In the enterprise data center, if you're talking linux, you're generally talking RedHat. That's where the ISV support is. RedHat no longer supports SPARC.
Second, most of Sun's x86 servers end up running Linux or Windows.
I've said this in other threads. Oracle gets deployed on Solaris/SPARC more than on Linux acco
Re: (Score:2)
As a matter of fact, I mostly agree with you. I just don't agree that there's any connection between Solaris's survival and the survival of Sun's x86 servers.
But yeah, the Oracle takeover is the best thing that could possibly happen to Sun's x86 products. Until now, they've been limited by the SPARC/Solaris Everwhere! mindset that still dominates Sun. (Better than it used to be — remember Cobalt? — but still pretty bad.) Now these servers will fall into the hands of a huge sales force that has n
Re: (Score:2)
Last reported quarter, Sun had around $916 million dollars in UltraSPARC/CoolThreads related hardware billings and only $176 million in x86 servers. The SPARC servers are supposedly also higher margin than the x64 servers.
SPARC sales are declining, while X86 servers (and CoolThreads) are increasing, but that's still a lot of money for SPARC. Add to that whatever the share of the $900 million in total hardware/os support fees that go towards SPARC based servers.
Just on those numbers alone, if you bought Sun
Re: (Score:2)
I mostly agree with your analysis (the SUV analogy is particularly apt) but I think you're making a very flawed assumption: that Sun's share of the x86 server market is as big as its going to get. As I've already argued, Oracle is in an excellent position to grow that market drastically.
Re: (Score:2)
very flawed assumption: that Sun's share of the x86 server market is as big as its going to get.
I'm not making that assumption. I think the x86 market will grow for Sun and I'm really surprised they haven't seen greater growth in that line. They have good designs and on their lower end, are price competitive with Dell. They even offer some larger servers and their relationship with Intel and AMD help them get them out to market fast. They were the first to come out with an 8 way quad opteron server. The x4600 [youtube.com]. And look. He picked up the server by the back of the chassis and didn't start bleeding :) O
Re: (Score:2)
To add. I also thing Sun has been getting a very raw deal.
Everyone is painting IBM as such the Linux supporter, but when you get IBM salespeople in, AIX and DB2 come up in the conversation.
IBM and HP have had Solaris to Linux migration campaigns, but not AIX or HPUX to Linux migration drives and if you search for AIX vs IBM you'll find some intersting stuff on IBM's site.
Solaris was (maybe still is) the leading Unix. HP and IBM didn't say Linux was better than Unix. They said Linux was better than Solaris a
Re: (Score:2)
Ah here is the sensible reply I was looking for. Now onto speculation:
As an example Oracle9i runs fine on a block device instead of a fs. The OS only seems to get in the way in the eyes of Oracle DB. That lends credence to the idea that Solaris vs. Linux is not worth the extra dev and maintenance costs. After all Oracle Corp already can sell you a turn key solution involving x86 boxes, Linux, and 10g. One thing that people running Oracle DBs on Linux wish they had was truss and dtrace though. Oracle Corp ma
Re: (Score:2)
> As an example Oracle9i runs fine on a block device instead of a fs. The OS only seems to get in the way in the eyes of Oracle DB.
Not entirely true.
1. Running Oracle on a block device instead of a filesystem makes certain classes of backups (like coldsnaps) a real pain in the ass
2. Ever since libaio, the filesystem layer is no longer significant in terms of Oracle performance (at least on Solaris, don't know about others).
ZFS with automatic hotsparing blah blah may actually improve things for filesystem
Re: (Score:2)
My brain boggles that you got AIO, VxFS, and Oracle DB to play nicely, wow you don't know how lucky you were or you did it under a certain mix of versions that was relatively stable.
I agree in point 1, but the performance increase was typically 30% or so. Thus that pain had to be dealt with.
The ZFS point, again why I see Oracle Corp seeing a place where there is money to be made.
Re: (Score:2)
Or, use a zvol instead of a full-fledged zfs filesystem. You get all the benefits of the storage pool and easy snapshotting, as well as all the benefits of a "bare blockdevice".
Re: (Score:2)
Does zvol know where there is/isn't data?
If not, snapshots would have to be the same size as the total storage available, which might get .. challenging .. in some cases.