Sun Microsystems came in just under the wire with its Solaris 10 10/08 update, squeezing it out on the last day of the month. The company had intended to do the launch a few days ago, but pushed it out to the other side of its Q1 financial results. The 10/08 update is not a major release, something that Sun has not put into the …
All Aboard the Gravy Train........ for the Turing Trip .... 42EdutainmentEntertainment.
"The predictive self-healing features" .... aka FailSafe Future Perfect.
I KId U Not. .....
"And if you did, you would probably want to kill yourself because you wouldn't be able to get anything done." ..... http://www.theregister.co.uk/2008/09/23/opensolaris_meet_solaris/ ... ITur2die4, Mr Roberts? Solaris 42 will allow for IT Growth with ITs Seamless Future ProgramMIng......by Virtual Machines/Alien Beings/Beta Humanity?
ITs Quantum Drive Operating System is the Alien Gift Offered to AI Sun GODs .... AIR&dD for/from GOD ... Advanced IntelAIgent Research and digital Development for Global Operating Devices.
Real Heavy Fuel ........and AI Quantum Leap into Virtual Processing rather than any Binary Firmware Evolution.
Perfect Sense? Ca Ira.
Please send me your CV, I predict a few vacancise will be opening up here at Sun soon, what with us having to let the whole chip dev team go. But we will have support for Intel's Nehelly- Newelly-... well, whatever it's called. Our tech guys have codenamed the new Intel stuff "The Only Viable Choice" for some reason, don't you just dig those crazy engineering minds? Anyway, I was very impressed with your eloquence and I'd love to see you head up our new Evangelising Depart- sorry, I meant to say Marketting Depratment. Anyway, drop me an email!
Hugs and kisses (well, virtual ones anyway, seeing as I don't physically interact with lesser beings, and virtualisation is the new Participation thing!),
Jonathan (still "God" to you even if I do give you a job).
ZFS for root is the killer feature
amanfromMars, you're making about as much sense as a certain Matty these days...
On another subject, the ZFS for root is awesome! You can create a liveupgrade ABE in seconds using a snapshot. Patch or upgrade the ABE without taking the system down. Boot into the new ABE. If you like it then make it active. If you don't or run into trouble, then boot off the original BE.
That is all.
This release will be the beginning of the end...
... for UFS and VxFS.
Zone patch on import
You don't mention what's probably the most interesting feature: until now it was only possible to transfer Solaris containers between systems which were at the same patch level; now you can import a container to a system which has patches at a higher level, and the patches will be applied to the container as it's imported.
This is a big deal. The ability to detach and clone Solaris containers is what really breaks the "one app, one system" bond, and instead enables systems to be used as capacity for containers. The reduction in numbers of systems when "test and dev" can be a few boxes for cloned containers rather than reference replicas of each production server is massive.
ROCK is the first cost reduction
Something is a miss a the circle K
Why does this ROCK link come up all the time?
If ROCK is dead as a dodo, why did the just put code into Opensolaris?
The article mentioned that this Solaris release was not a major one, well I consider this release being a major milestone :)
ZFS root, more dtrace capabilities and the "zone patch import" mentioned above make this release very very important.
RE: Zone patch on import
Whilst it sounds an interesting idea, it is not one I would implement as anything other than a test feature. I have seen plenty of cases with several OS where adding a later patch has broken an app. It also highlights one of the great advantages of hp-ux and Integrity Virtual Machines - unlike containers, an Integrity Virtual Machine is a complete and separate OS instance, so the patch level is irrellevant. I can (and do) have three development itterations at different patch levels on the same hp-ux partition, in separate VMs. And I can take the boot disk LUN from any of those VMs and import it into a new VM on a completely different server hosting Integirty Virtual Machines without having to worry about any patch changes due to differences in the host system. More to the point, our dev systems can host multiple patch versions of multiple different OS (Windows, hp-ux, Red Hat Linux, and VMS) all on the same system, a trick which Sun can only dream of. So when it comes to making the best return on investment on a dev system, Sun and Slowaris are still miles behind.
As long as Matt (HP Marketing), and the IBM guys we had on the other article, plus whichever other clever astroturfers (including the owners of the anonymous geocities pages) are still taking the time to bash Sun out of their hectic schedules it seems Sun are still a worry to them.
Perhaps the time to worry is when we have a Sun article and no backbiting from other vendors, just dismayed fanboys?
RE: Zone patch on import
Matt, you are not really comparing the same technologies, bring in LDOM, VirtualBox or xVM and then compare, also two of your 4 OS's are obsolete in todays thinking so I would not shout about that from the rooftops.
I agree I would not import a production zone onto a different host at a different patch level unless it was part of an upgrade I was planning but the facility to move zones around and onto new hosts with a higher patch level is very compelling and looking at the web has been requested by many people.
RE: zone patch on import
As Matt says, zones are not complete OS instances. LDOMs (on Niagara and related processors) provide that, and on x86 there are obviously a lot of solutions now. All of these provide much higher levels of isolation.
Zones do have significant advantages however. An important one is that they are much more "inspectable" than a VM.
It makes me laugh, seeing you Sun guys getting all excited about features which have been around for ages in AIX.
Live upgrade of an ABE to clone the running OS, and patch it. Great. This could be done over 2 years ago on AIX using multibos.
As for LDOMS. they are at least 8 years behind AIX in terms of features. Dynamic memory and I/O re-allocation anyone ??
Re: ZFS Boot
LDOMS are 8 years behind dynamic memory and I/O on AIX? Isn't that just dynamic reconfiguration which is available on Sun Enterprise servers?
Live Upgrade has been available in Solaris much longer than 2 years! (since Solaris 9)
Using ZFS just means that the new BE (boot environment) is available in a few seconds (by creating a writable snapshot - clone) instead of a hour as it was in UFS (by creating a disk to disk copy of the filesystems).
Does AIX's multibos allow you to create a clone of the BE in seconds?
Yes, a running clone of your boot environment. An OS within an OS if you like, and takes just a few minutes.
The clone environment can be patched, re-configured etc without touching the running system. Just a reboot to activate, and leaves the existing environment intact for fast back out, if required. Works a treat.
"and takes just a few minutes."
So you can completely copy your entire boot environment in a few minutes? Yeah right... What's a few to you? 20 minutes? 60 minutes? Come on multibos is relatively new technology when compared to liveupgrade (around for at least 8 years). Creating a liveupgrade image in a ZFS snapshot is very cool. IBM will probably try to copy it, as they have most other things from Solaris,
but that's OK, as Solaris has been trying to copy the Mainframe for years...
Yes, a few minutes is fairly typical. I guess if you have large logical volumes in rootvg, it can take a little longer, but then rootvg should be kept lean anyway.
Multibos has also been available for 2-3 years. ZFS boot environment cloning only out just now.
Btw, AIX has also had alt_disk_install since v4.3 which being 10 years old pre-dates your Solaris 8 live upgrade.
Face it, all vendors copy one another. What next ? Accusations of IBM copying LPARs from LDOMs !