Hardly unnecessary bitterness, more like relevant points made for considering other options, and seeing that it's coming from the people that worked closely on the stuff it's pretty darn revealing.
Funniest moment ever: the Illumos leadership reasoning that, as Sol11 is too much change for Sol10 shops, then Illumos has an opportunity to enter these places as an alternative. Right, Illumos is SO much like Sol10 that they won't even notice.
Illumos is OpenSolaris 2.0.. same people, same mentality, how could things be any different?
"Funniest moment ever: the Illumos leadership reasoning that, as Sol11 is too much change for Sol10 shops, then Illumos has an opportunity to enter these places as an alternative. Right, Illumos is SO much like Sol10 that they won't even notice."
And they are correct! Solaris 11 is NOT an option because of IPS and because of lack of Flash(TM).
An Illumos-based distribution based on SVR4 packaging which supports Flash(TM) would be a viable alternative to Solaris 11. And considering that Illumos regularly gets fixes and enhancements to the source code, it would be a worthy alternative.
Why, Illumos even has driver support for hardware which Solaris 11 does not even support!
Not only that, but Oracle has stated in no uncertain terms, that they will not compete in the i86pc space.
That in turn means that Solaris is going the way of AIX, it will only support Oracle hardware. As time passes the i86pc architecture will evolve, and Solaris will not support the new hardware beyond that which will be found in "Snoracle" hardware.
UX-admin: while some recent Oracle decisions are questionable I don't think there is a need for FUD. They do support 3rd party x86 hardware and they do add 3rd party x86 hardware to HCL. There are also new drivers and updates to old drivers in recent builds not to mention fixes for the S11 Express, and yes for x86 too. And yes, there are regular updates to S11 express and also there are still bi-weekly dev builds being produced.
While AI is still lacking some features the truth is you can get it working as I have. No flash archives - well, while this is true IIRC from Caiman mailing list this is planned in some form. Now show me Illumos based distro with a flash archive support and no IPS...?
The truth is that currently there is no full blown enterprise ready distro based on Illumos. I hope there will be one one day but we are not there yet. While there is already Solaris 11 Express and FCS is around a corner. And for the near future at least Solaris 11 is the way to go in enterprise both on SPARC and x86.
Instead of spreading FUD and fighting each other we will be much better of by co-operating, even if Oracle isn't much of a community player right now, imho.
"UX-admin: while some recent Oracle decisions are questionable I don't think there is a need for FUD. They do support 3rd party x86 hardware and they do add 3rd party x86 hardware to HCL. There are also new drivers and updates to old drivers in recent builds not to mention fixes for the S11 Express, and yes for x86 too. And yes, there are regular updates to S11 express and also there are still bi-weekly dev builds being produced."
So I saw. However, this is no FUD: Oracle, Larry Ellison himself in fact has states on almost every earnings call that "Oracle is not in the business of x86 servers and we will not compete in that business". Those are his statements, not mine.
And based on that, as well as lack of advertising and minimal amount of new i86pc servers from Snoracle, while sparc is being heavily advertized, pardon me but I do not see how the next generation Solaris will support new consumer-grade electronics, when the focus is clearly on servers, what is left of them in the i86pc arena.
How??? Logically, based on the information available, how???
They already ditched 32-bit, which rules out lots of mini-ITX servers based on AMD and intel.
"While AI is still lacking some features the truth is you can get it working as I have. No flash archives - well, while this is true IIRC from Caiman mailing list this is planned in some form."
That's great, but I need Flash(TM) yesterday! And "in some form" implies incompatibility with existing flar commands and format(s).
We have significant investment in Flash(TM) technology, and it would be irrational to throw it all away because Bart Smaalders & Co. feel they can reinvent the wheel better.
Thanks Bart & Co., but no thanks. Do not want your technology. I am not about to eat the financial cost of your experiments, and neither is my company.
"Now show me Illumos based distro with a flash archive support and no IPS...?"
SchilliX. It does not have Flash(TM), but it is SVR4 based, which is extremely important.
"Instead of spreading FUD"
Don't accuse me of something that I am not doing. You yourself know how bad the situation with Snoracle is right now, and denying it only makes it worse.
Solaris 11 is a losing proposition, primarily because of "innovations" like IPS and AI.
All the major innovators of Solaris have left and are working on Illumos now. And you know it. So do I.
"and fighting each other we will be much better of by co-operating, even if Oracle isn't much of a community player right now, imho."
We cannot rally together, as we are divided on IPS.
"I do not see how the next generation Solaris will support new consumer-grade electronics, when the focus is clearly on servers"
I thought we were talking about servers here and not toys for home. Although it would be nice to be able to run Solaris on both - and perhaps thanks to Illumos it will be possible.
"They already ditched 32-bit, which rules out lots of mini-ITX servers based on AMD and intel."
Frankly, I don't care. Even my home server has a 64bit AMD CPU (and yes, it is running Solaris). If Illumos wants to keep support for 32bit CPUs and there is demand for it - great.
"SchilliX. It does not have Flash(TM), but it is SVR4 based, which is extremely important."
I'm sorry but SchilliX is nothing more than a toy. Comparing Solaris 11 and SchilliX is just pointless. One provides a full blown OS with global commercial support the other provides neither.
And what's wrong with IPS? I actually like it and it is a great improvement over the old packaging system.
Lack of image based installations - as stated before it is on their roadmap. And you can still use Solaris 10 for many years to come. Other than that afaik nor SchilliX or Open Indiana provide it neither.
btw: while I like IPS I don not actually like AI. Having said that it does its job at the end.
"I thought we were talking about servers here and not toys for home."
How does one drive adoption if people can not play with it at home?
How does one spread Solaris to embedded platforms, many of which are 32-bit, if support for it has been ripped out?
Besides, USB, scsa2usb(7D) in particular, never worked correctly and scsa2usb(7D) still to this day does not work correctly even in the latest incarnations of Solaris 10 and Solaris Express.
Now the chance that it ever will is close to none.
But, let us keep discussing servers: now that only Snoracle and hp (and maybe DELL) hardware is supported, what will happen to shops which run Solaris on TYAN, Supermicro, intel, Silicon Mechanics, and the like?
"Frankly, I don't care. Even my home server has a 64bit AMD CPU (and yes, it is running Solaris)."
I have a farm of perfectly good, last-generation 32-bit Pentium 4 3.2GHz ASUS 1U servers. There is nothing wrong with them. But now, the kernel will not support them.
"I'm sorry but SchilliX is nothing more than a toy. Comparing Solaris 11 and SchilliX is just pointless. One provides a full blown OS with global commercial support the other provides neither."
I am a system engineer, by profession, and am not afraid to modify kernel code and thus, I require no support. In 15 years that I ran Solaris, I only opened two support calls. One had to do with scsa2usb, the other with lp(1).
So support is clearly not a factor for Bryan Cantrill, nor for Adam Leventhal, ... nor for me.
The more advanced the shop, the less chance of having a support contract.
Like I wrote earlier, I have engineered my own Solaris distribution before, and I will do so again if need be. I prefer not to because it costs money, but I would rather do that than run Solaris 11 and IPS.
SchilliX might be a good place to start. Joerg usually knows what he is doing, much to everybody else's dismay, but not mine.
"And what's wrong with IPS?"
It does not support scripting. One can emulate postinstall via SMF; there is no emulation for preinstall; and since IPS isn't tied into SMF, there is absolutely no way to emulate preremove and postremove via SMF.
Why is this so important? Because if your servers have configuration performed via packages, IPS is not a viable solution. By design. That is severely BUSTED, especially if we consider that SVR4, swinstall, inst, and RPM can do it.
Hardly unnecessary bitterness, more like relevant points made for considering other options, and seeing that it's coming from the people that worked closely on the stuff it's pretty darn revealing.
ReplyDelete+1 Dave..Really...
ReplyDeleteFunniest moment ever: the Illumos leadership reasoning that, as Sol11 is too much change for Sol10 shops, then Illumos has an opportunity to enter these places as an alternative. Right, Illumos is SO much like Sol10 that they won't even notice.
ReplyDeleteIllumos is OpenSolaris 2.0.. same people, same mentality, how could things be any different?
"Funniest moment ever: the Illumos leadership reasoning that, as Sol11 is too much change for Sol10 shops, then Illumos has an opportunity to enter these places as an alternative. Right, Illumos is SO much like Sol10 that they won't even notice."
ReplyDeleteAnd they are correct! Solaris 11 is NOT an option because of IPS and because of lack of Flash(TM).
An Illumos-based distribution based on SVR4 packaging which supports Flash(TM) would be a viable alternative to Solaris 11. And considering that Illumos regularly gets fixes and enhancements to the source code, it would be a worthy alternative.
Why, Illumos even has driver support for hardware which Solaris 11 does not even support!
Not only that, but Oracle has stated in no uncertain terms, that they will not compete in the i86pc space.
That in turn means that Solaris is going the way of AIX, it will only support Oracle hardware. As time passes the i86pc architecture will evolve, and Solaris will not support the new hardware beyond that which will be found in "Snoracle" hardware.
Thanks, Oracle, but no thanks.
UX-admin: while some recent Oracle decisions are questionable I don't think there is a need for FUD. They do support 3rd party x86 hardware and they do add 3rd party x86 hardware to HCL. There are also new drivers and updates to old drivers in recent builds not to mention fixes for the S11 Express, and yes for x86 too. And yes, there are regular updates to S11 express and also there are still bi-weekly dev builds being produced.
ReplyDeleteWhile AI is still lacking some features the truth is you can get it working as I have. No flash archives - well, while this is true IIRC from Caiman mailing list this is planned in some form. Now show me Illumos based distro with a flash archive support and no IPS...?
The truth is that currently there is no full blown enterprise ready distro based on Illumos. I hope there will be one one day but we are not there yet. While there is already Solaris 11 Express and FCS is around a corner. And for the near future at least Solaris 11 is the way to go in enterprise both on SPARC and x86.
Instead of spreading FUD and fighting each other we will be much better of by co-operating, even if Oracle isn't much of a community player right now, imho.
"UX-admin: while some recent Oracle decisions are questionable I don't think there is a need for FUD. They do support 3rd party x86 hardware and they do add 3rd party x86 hardware to HCL. There are also new drivers and updates to old drivers in recent builds not to mention fixes for the S11 Express, and yes for x86 too. And yes, there are regular updates to S11 express and also there are still bi-weekly dev builds being produced."
ReplyDeleteSo I saw. However, this is no FUD: Oracle, Larry Ellison himself in fact has states on almost every earnings call that "Oracle is not in the business of x86 servers and we will not compete in that business". Those are his statements, not mine.
And based on that, as well as lack of advertising and minimal amount of new i86pc servers from Snoracle, while sparc is being heavily advertized, pardon me but I do not see how the next generation Solaris will support new consumer-grade electronics, when the focus is clearly on servers, what is left of them in the i86pc arena.
How??? Logically, based on the information available, how???
They already ditched 32-bit, which rules out lots of mini-ITX servers based on AMD and intel.
"While AI is still lacking some features the truth is you can get it working as I have. No flash archives - well, while this is true IIRC from Caiman mailing list this is planned in some form."
That's great, but I need Flash(TM) yesterday! And "in some form" implies incompatibility with existing flar commands and format(s).
We have significant investment in Flash(TM) technology, and it would be irrational to throw it all away because Bart Smaalders & Co. feel they can reinvent the wheel better.
Thanks Bart & Co., but no thanks. Do not want your technology. I am not about to eat the financial cost of your experiments, and neither is my company.
"Now show me Illumos based distro with a flash archive support and no IPS...?"
SchilliX. It does not have Flash(TM), but it is SVR4 based, which is extremely important.
"Instead of spreading FUD"
Don't accuse me of something that I am not doing. You yourself know how bad the situation with Snoracle is right now, and denying it only makes it worse.
Solaris 11 is a losing proposition, primarily because of "innovations" like IPS and AI.
All the major innovators of Solaris have left and are working on Illumos now. And you know it. So do I.
"and fighting each other we will be much better of by co-operating, even if Oracle isn't much of a community player right now, imho."
We cannot rally together, as we are divided on IPS.
And something else: I have enough knowledge to build my own Solaris distribution, right down to my own JumpStart(TM) compatible installer.
ReplyDeleteI have already done that with Solaris 10, so a distribution based on Illumos is not out of the question.
So I do not really depend on someone else's good grace and mercy to release the kind of distribution that is a worthy sucessor to Solaris 10.
And obviously, I have very high esteem for Solaris 10 and technologies contained within. There is a reason for that.
"I do not see how the next generation Solaris will support new consumer-grade electronics, when the focus is clearly on servers"
ReplyDeleteI thought we were talking about servers here and not toys for home. Although it would be nice to be able to run Solaris on both - and perhaps thanks to Illumos it will be possible.
"They already ditched 32-bit, which rules out lots of mini-ITX servers based on AMD and intel."
Frankly, I don't care. Even my home server has a 64bit AMD CPU (and yes, it is running Solaris). If Illumos wants to keep support for 32bit CPUs and there is demand for it - great.
"SchilliX. It does not have Flash(TM), but it is SVR4 based, which is extremely important."
I'm sorry but SchilliX is nothing more than a toy. Comparing Solaris 11 and SchilliX is just pointless. One provides a full blown OS with global commercial support the other provides neither.
And what's wrong with IPS? I actually like it and it is a great improvement over the old packaging system.
Lack of image based installations - as stated before it is on their roadmap. And you can still use Solaris 10 for many years to come. Other than that afaik nor SchilliX or Open Indiana provide it neither.
btw: while I like IPS I don not actually like AI. Having said that it does its job at the end.
"I thought we were talking about servers here and not toys for home."
ReplyDeleteHow does one drive adoption if people can not play with it at home?
How does one spread Solaris to embedded platforms, many of which are 32-bit, if support for it has been ripped out?
Besides, USB, scsa2usb(7D) in particular, never worked correctly and scsa2usb(7D) still to this day does not work correctly even in the latest incarnations of Solaris 10 and Solaris Express.
Now the chance that it ever will is close to none.
But, let us keep discussing servers: now that only Snoracle and hp (and maybe DELL) hardware is supported, what will happen to shops which run Solaris on TYAN, Supermicro, intel, Silicon Mechanics, and the like?
"Frankly, I don't care. Even my home server has a 64bit AMD CPU (and yes, it is running Solaris)."
I have a farm of perfectly good, last-generation 32-bit Pentium 4 3.2GHz ASUS 1U servers. There is nothing wrong with them. But now, the kernel will not support them.
"I'm sorry but SchilliX is nothing more than a toy. Comparing Solaris 11 and SchilliX is just pointless. One provides a full blown OS with global commercial support the other provides neither."
I am a system engineer, by profession, and am not afraid to modify kernel code and thus, I require no support. In 15 years that I ran Solaris, I only opened two support calls. One had to do with scsa2usb, the other with lp(1).
So support is clearly not a factor for Bryan Cantrill, nor for Adam Leventhal, ... nor for me.
The more advanced the shop, the less chance of having a support contract.
Like I wrote earlier, I have engineered my own Solaris distribution before, and I will do so again if need be. I prefer not to because it costs money, but I would rather do that than run Solaris 11 and IPS.
SchilliX might be a good place to start. Joerg usually knows what he is doing, much to everybody else's dismay, but not mine.
"And what's wrong with IPS?"
It does not support scripting. One can emulate postinstall via SMF; there is no emulation for preinstall; and since IPS isn't tied into SMF, there is absolutely no way to emulate preremove and postremove via SMF.
Why is this so important? Because if your servers have configuration performed via packages, IPS is not a viable solution. By design. That is severely BUSTED, especially if we consider that SVR4, swinstall, inst, and RPM can do it.