IBM's AIX 6 drops 'L,' adds 'S' Goodbye, AIX 5L Version 5.4. Hello, AIX 6 – soon shipping with less L and more S. Just a couple days after revealing its first Power6-based servers, IBM has handed customers an outline of the expected features in AIX 6 – or the operating system formerly known as AIX 5.4. The Unix OS, IBM confirmed this week, will enjoy a public …
"IBM apparently ... will introduce the software as AIX Version 6.1"
I understand that some marketing droid committee in IBM corporate has decided that customers are too easily spooked by ".0" releases, so in order to avoid that ... there won't be any in future. Us customers are so easily fooled that we won't realise it's a major new release, we'll be lulled into a false sense of security by the fact that it's ".1", not the dreaded ".0".
I first heard this from a product manager who told us that the next version of the product we use would be "11.1", when the current version was "10.2". When I asked 'what happened to "11.0"?', they told the audience that 'this was a marketing decision'
?
marketing guys never quite know exactly what they are selling, do they ?
the explanation is quite simple.
since the beginning of aix, x.0 versions are specifically meant for early adapters, evaluators; e.g, programmers testing their software on compatibility etcetera.
these x.0 versions were never upgradeable to x.1 and beyond, you had to reinstall. not what you want for your producton environment.
hence the 6.1 , probably to avoid an avalanche of questions from AIX knowledgeable persons.
PS
the wpar is indeed the same as zones. the mobility is something that
SUN/HP should have a look at. ( I've seen a beta version working one year ago, btw, on aix 5.3 with oracle 10. )
there will also be lpar mobility, that is, moving the entire OS ( global zone for solaris people ) live, without service interruption. I have seen that working too, beta-version of course.
If you look on the IBM site you will find that IBM run a number of forums with customers for collecting requirements. All of the above were at the request of customers - rather than copying Sun.
As far as I know Sun doesn't do LPAR's, so the WPAR and partition mobility facilities were part of a large partitioning strategy that was first seen in zSeries/s390 mainframes long before Sun introduced zones/containers. These facilities are to manage upgrades and resources better.
The story going the rounds at the moment is that the beta you sign up for will be for "AIX 6.0" and that's why the production release is "AIX 6.1". Okay, I suppose it doesn't hurt that this neatly avoids the ".0 releases suck" perception out there.
As to the "containers", sorry "Workload Partitions", yes it's a good idea and if it adds to the flexibility of the system then I've got no objection. Maybe Sun can return the favour by "borrowing" the Mobility idea for their implementation.
As to the other stuff "borrowed"(?) from Solaris - well it's "embrace and extend"...
... oh heck, I just really hope that this doesn't mean that AIX 7 will be rebranded as "AIX Vista"! :-P
If AIX encrypts the data in the filesystem, at source. Will a de-duplication device still be able to detect duplicate blocks of data and reduce the backup window. Or will de-duplication stop working, will de-duplication have to decrypt it before backing up, the encrypt again when storing the backup copy. What is the overhead and security exposure of this.
More details here: http://blogs.sun.com/ValdisFilks/entry/the_dupe_in_de_duplication
Execellent article, however, the proctologists section was painful, not sure if I want to delve further into this area.
Just want to correct one point in the comments above;
<snip>"the wpar is indeed the same as zones. the mobility is something that SUN/HP should have a look at. ( I've seen a beta version working one year ago, btw, on aix 5.3 with oracle 10. )" </snip>
<snip> As far as I know Sun doesn't do LPAR's, so the WPAR and partition mobility facilities were part of a large partitioning strategy that was first seen in zSeries/s390 mainframes long before Sun introduced zones/containers. These facilities are to manage upgrades and resources better. </snip>
Not sure where this information has come from, as in one of our environments we've been running numberous containers/local zones across a large number of frames (think over 100), and a lot of testing and work has been done not only in the reallocation of resources between domains and containers, but also in moving the entire zone/container. This works fine from domain/partition to domain/partition, and is still relatively seamless if you're moving to a completely different frame! Zones can be moved, domains can be moved. SUN doesn't need to 'borrow' mobility from anyone 'cause it's already there!
I'm just wondering when HP are going to jump on the bandwagon. Besides, it's a good 'excuse' for everyone to be greener - let's run our whole datacenter on two or three bits of (frightfully) expensive kit... Just make sure your UPS's and cooling are set up properly (as a certain Australian location found out).
So far, having had some exposure to Sol 10.x and having upgrades to AIX 6.x on the way, it'll be interesting to see how much is actually different apart from the obvious implementation method...
If IBM Marketing bonuses are dependant on reduced sails.
The wierd numberibng schemes. The bizarre branding
( The ancient OS, OS/MVS ..... zOS copy utility written circa 1962 is now a branded a Tivoli product ).
Succesful products like the mainframe operating system with 16 names are contantly and pointlessly renamed, while hated products (You know who you are Lotus Notes) are lumbered with the old branding even after they have improved enough not to produce groans of despair from the users.
Then there are product names/numbers:- sales have wizzy names for products, but when you come to buy something you have to navifgate a different and seamingly unrelated set of products codes, then, after you've bought it you need to supply support with some wierd 9 digit part number which goes back to the days when punched cards ruled.
Great products - bizarre marketing.
HP-UX has had soft partitioning, hard partitioning, and any number of combinations in between, for ages. Work Load Manager for HP-UX has been letting multiple apps run without infringing on each other in the same OS instance under HP-UX for so long I can't remember when it was intro'd (I think it was around 2000?), and GlobalWLM has long allowed the balancing of apps across multi-OS systems. Sorry, but IBM are sooooo far behind on this. And, more importantly, HP-UX has it all with an integrated management suite which not only manages virtualisation of multiple multi-OS instances but also their capacity-on-demand technology and GlobalWLM from one pne of glass, and this can then feed upwards into HP OpenView Operations for a complete picture of your environment by system or service view. IBM wish they had that kind of capability.
And switching an instance between nodes running Oracle 10g.... erm... isn't that what Oracle 10g is all about?
HP have all this without forcing another CPU architecture (Power6 looks like fast-but-dumb Power5, like T1 is superfast-but-superdumb SPARC) and a new version of the OS (what is binary compatibility between AIX 5L and 6.1?) and a recompile of all the apps ('cos dumb-but-fast Power6 requires no out-of-order code). Call me an HP troll if you like, but come on, Ashlee, try reading up on HP-UX, puh-lease!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
