... I don't want to use stinking AIX.

Last time I quit the job as management decided to move to POWER/AIX.

More important than the hardware, is still the operating system.

You can pry (Open-)Solaris from my cold, dead hands.

IBM and HP can probably afford to tighten their belts and offer big discounts in the hope of coralling future business - if Sun tightens its belt any more they'll cut themselves in half. Just like the post-Y2K downturn, it's a great time to play one vendor of against another to get a bargain!!! The last time round it was storage, IBM was targeting EMC accounts, and we got a Shark free for a year. IBM gave it to us with free support in the belief we were going to use it for mission-critical data, obviously hoping that once we'd got important data on it we'd be too scared to back it out, but in the end we just used it for backups. I still smile when I remember the look on the IBM salegrunt's face when we said we were not keeping it! Anyway, I'm off to see what UltraSPANKED kit we have left stored in the basement that we can get IBM and HP to fight over......

You can keep your Open Solaris.

Personally, I am more than happy with the super stable, feature rich AIX & Power systems, which imho is far superior combination than anything Sun has to offer right now.

but of course, opensolaris on power6 would be very cool, no ?

if IBM ported & supported an opensolaris implementation, they'd probably have a lot of people starting to line up to buy

I have worked with nearly every UNIX flavour and I just dont understand why Solaris has so many fans.

Its not bad, certainly not as bad as HP/UX but not nearly as good as AIX or the old DEC UNIX.

AIX is the most POSIX compliant unix so its the easiest base to port C/C++ programs to and from.

System management in AIX is just the best. It takes about two minutes to hook up and configure some new storage using "smitty", the same operation on Solaris takes about three hours (including an hour trying to locate the right manual to RTF).

To anyone SUN shops who are thinking of taking IBMs thirty pieces of silver: you will never regret it, but, it would be wise to check how the move affects your ORACLE licenses first. Larry Ellison has too much money already.

Is it not true that it is possible to reduce Oracle licensing costs, as Power has faster cores, meaning you can run same workload, on less cores = less licenses.

Well I think that both Solaris and AIX are the 'good guys' when it comes to following POSIX, You might argue that AIX typically uses the POSIX values by default and you sometimes have to specify it with Solaris, not like HPUX *SIGH*.

I do think that Solaris and AIX are a bit different. Solaris being the favourite of the app. developers, and AIX being the favourite of the Sysadmins. I had to debug some IO performance problems on a Solaris 9 box some years ago. And I was amazed at how primitive the tools were compared the the ones I was used to from AIX 5.3.

And when you talk to AIX and Solaris fanatics you usually notice that the AIX fanatics are almost all Sysadmins, but the Solaris fanatics have a very high percentage of DB admins and developers among them.

The reason that I think that 'AIX servers' are the best, is the combination of hardware, virtualization layer (POWERVM) and the OS.

Which means that IMHO there really aren't any clear weaknesses when you compare an AIX solution to one based upon Solaris or HPUX.

With regards to Oracle licensing, If you move from SPARC to POWER6, then you'll end up with your licensing savings paying for your new servers, if you do it right. It takes a little design work to harvest the savings but the POWER hardware, POWERVM and AIX will give you the tools to do it.

Sure you have to fight off Oracle sales people (or some other software company) who will try to trick you, cause they don't like loosing licensing fees.

I once had an Oracle sales guy who wanted to have licensing fee for x3 the number of cores in the system, because we used uncapped micropartitions. But luckily the Oracle licensing are pretty clear.

And remember there is NO EXTRA VALUE in paying for 12 licenses compared to paying for 5 if they are of the same type. But having better hardware does give you value.

And just to close the Oracle bit. Do not.. do not.. not not not run Oracle on the T5X40 servers, they make amazing (but far to expensive) webservers, but license cost will ruin you.

// Jesper

My view is that Sun virtualization offerings are quite poor in comparison to IBM. Ok, zones/container technology is maturing, and has its place, but is not always suitable. Of course, IBM now has WPARs which is essentially the same thing, and actually easier to configure, and has more features.

LDOMs are more akin to Power 4 technology, which goes back over 8 years, and are light years behind the capabilities of IBM VIO & Power 6. I would argue that LDOMs are not even true virtualization, more like thead partitioning.

Factor in that Sun virtualization only runs on limited hardware, and only on Solaris 10, whereas IBM runs across the full range. No wonder why Sun are losing market share.

I've used AIX and Solaris. I haven't used HPUX too much, but the folks that I talk to that do use HPUX don't sing its praises. HPUX users just like HP, and HPUX just comes with the bargain. As far as Solaris 9 versus AIX 5.3, I agree, there are some missing management pieces. As far as Solaris 10, AIX is years behind. Solaris 10, once you get used to some of the new pieces is so much manageable that anything else out there it's embarrassing. Except patch management... I hope the new tools coming out with ZFS and Liveupgrade will live up to the hype.

the newest AIX version is AIX 6.1 not 5.3.

ZFS looks nice but many of the features I see pointed out have been in AIX for years Snapshots compressed filesystems or have been added in AIX 6.1 cryped filesystems etc.

Not that ZFS doesn't seem more advanced than JFS2 cause it does.

As for patch management then AIX have always been good at that. I mean I accidental patched some key libs that a running Oracle database used many years ago, now that is bad. But after just doing a rollback of the patches I could actually close down what should be closed down and repatch the system.

I also remember being able to backup a filesystem where there weren't any disks below it, straight from filecache. Saved a whole development departments from some red ears.

But still it's the whole package that counts, and much of the work done on 5.3 and 6.1 seems to have gone into virtualization, or rather to integrate AIX with the virtualization done by POWERVM.

I mean they must have rewritten a lot of code, just the accounting part must have been a nightmare. Imagine you have 100 virtual machines on a machine using 200 virtual processors that run ontop of 32 actual cores, and all those virtual machines all run at more than 100% utilization if you use standard sar commands. Brrr...

// jesper

You dont necceseraly end up forking out a bigger wodge to Oracle when you switch but you should be careful. Oracle licensing terms are classic of Dilbert "confusopoly" and weasel wording you can get stuck with a very large bill for not much database if you dont read the small print.

To reiterate my previous point its not that solaris is a bad operating system. Its just not so much better that you should put up with inferior hardware just to stay on solaris.

I personaly prefer AIX as a development platform. The only caveat there is that power hardware is very forgiveing when it comes to word boundries, it will quite happily load an integer from an odd numbered address. Code that works perfectly on power hardware often falls flat on its face when ported to other boxes.

Care to back that statement up ?

Scott's Top 10 list

1) Sun's highend is in a free fall because customers are moving to Power rather than migrating to Fujtisu technology.

2) Sun just announced end of life (January) for USIV+, and the (Fujitsu) M class is not very profitable compared to their own highend

3) Sun only wrote down the good will of hardware acquisitions, when will they write down the MySQL $1B acquisition?

4) Suns MySQL revenue/growth is hidden with all the other middleware and the MySQL added.

5) JS may not hear customers say they are moving to Power, but that is because he is not talking to customers who are throwing them out for Power systems

6) How can JS talk about additional acquisitions several times when they are writing off billions from prior acquisitions as failures?

7) Sun's revenue is in free fall because the MySQL acquisition and Sun's sales strategy is to replace all Oracle installations with MySQL and they are at war in the sales force. What they miss is most Oracle customers have enterprise license agreements and MySQL customers on Power are very happy to run their business on Power.

8) Sun does not have a leader who inspires the team. Nicknames are a symptom of a disgruntled workforce

9) Page 6 is an indication that Sun is being more transparent so someone will buy them for the value of the parts.

10) The clock is ticking and there is no guidance for anything better.

http://www-03.ibm.com/systems/power/

"As for patch management then AIX have always been good at that. I mean I accidental patched some key libs that a running Oracle database used many years ago, now that is bad. But after just doing a rollback of the patches I could actually close down what should be closed down and repatch the system."

This is way too simplistic. Everything you talk about here has also been in Solaris for ages... You don't understand what Liveupgrade and ZFS bring to the table. They work in conjunction, you can use Liveupgrade to create an alternate boot environment on a ZFS snapshot. Patch the alternate boot environment. Wait for a convenient time to reboot and make it active (yes, boot a snapshot). Your only down time for patching is the reboot. If the new patches cause a problem reboot back into the original environment. If you like the new patched environment, then mark it as good and your good to go.

You can't do that in AIX or HP for that matter.

I'm working for an organisation with ~20k servers in Europe, the lion's share of them being Sun.

But, guess what? New acquisitions are all IBM servers. No new Sun kit is being deployed, and nobody at Sun seems to give a damn. As someone with three Sun certifications, this worries me slightly (it basically means my next certification will be from IBM, not Sun) - but it should be giving Sun management bed-wetting nightmares. Yet, it's not penetrated their skulls at all.

Funnily enough, I used to work at Sun - under McNealy - and he had the same core problem that Schwartz has: Too much ego, not enough listening to customers. The customers are, in turn, pissed off with Sun - and as a result, have moved IBM and HP to the top of the preferred vendor list. I don't think Sun management will realise the danger until they're in Chapter 11.

It's a major shame, but Sun just doesn't seem to realise (or appreciate) where its cash is coming from. In a time when it really needs to be pulling out the stops to ensure its customers are happy, it's actually driving people into the arms of IBM and HP because its own account management are so crap. I've witnessed several take stances that infuriated their customers, and even seen on-site Sun PS contractors fired because of this. Orders for new Sun kit also stopped at about the same time...

Whatever happened to the attitude "The customer is always right"?

Dear Bill.

The functionality you talk about there have been available in AIX since AIX 4.3 and is called Alternate disk install. Now let me point out to you that AIX 4.3 is more than 10 years old.

The Original feature let you clone the OS of a running system to a new set of disks and you could then patch do a full os upgrade whatever on the OS image on these disks, and then boot from them and then boot back to the original disks if you wished so. You could also use this feature to clone systems, and that could be useful for systems where you hadn't any network on them yet. Have had to do that quite a few times.

Now the current version of alternate disk install is quite a bit more advanced, and doesn't need disks to clone on.

I must admit I'm getting a bit tired of hearing smart fancy Solaris marketing names like Jumpstart, liveupdate for things that have been on AIX for years. Sure Network Installation Manager (NIM) isn't as fancy a name as 'jumpstart' and liveupdate sounds much much better than alternate disk install. But that is AIX, no catchy names and fluff. It's just a really really good UNIX, and under it you have a hypervisor that's perhaps only triumphed by zSeries, and under that again you have the damn best hardware you can get.

It's like the A10 warthog: ugly, effective and almost impossible to bring down.

It's not that Solaris is a bad OS, I'd just think that AIX is a little better. But if you factor the hypervisor and the POWER hardware into the equation. Then it's really not a race.

// Jesper

A couple years ago my management decided to switch from Solaris 8/9 and HP-UX 11i v1/2 servers to AIX 5.3 LPARs to save on Oracle licensing. Replaced 20 Sun and HP servers for SAP and Oracle with two 8-way IBM POWER5+ boxed running everything through VIOS and uncapped micropartitions, and still had plenty of CPU processing power. Now have 80 AIX 5.3/6.1 LPARs across two 8-way POWER5+ servers and two 8-way POWER6 servers for SAP, Oracle, UniData, and ESRI--everything using VIOS and uncapped micropartitions. (Also now have SAN volume controllers, and they are very slick, too.) I can't imagine how much we would have spent in Oracle licensing if we didn't do this.

And after using Linux, Solaris, and HP-UX for years, AIX took some time getting used to, but now I prefer to use it over any other UNIX. It is really a sysadmin's operating system.

Im afraid your quite wrong.

Cloning of a running AIX instance has been around for years. As Jesper points out, 1st with alt_disk_install, and with AIX 5.3 also has a feature called multibos, which means you can clone the running OS within the same root volume group. This means you dont even need to clone to different disks. The clone instance can be patched, re-configured, whatever you want to do with it, and then simply boot from it to make it active. The original instance remains intact and untouched, so you have a fast back out option, if needed.

OK, I think I'm going to be accused of bias in the opposite direction but lets have a go :

Solaris Liveupgrade (ie: create an upgrade image while the server is live) seems an adequate description, has been around for years, at least since Solaris 8 & that was out in 2000. OK, so AIX 4.3 beat it by a few years but neither technology could be accused of being immature.

Your MultiBos technology point is identical to that offered by live upgrade (ie: roll OS forward without affecting live, if you dislike rollback again with a command and a reboot).

Disk config is laughably easy to me so how you took three hours I don't know? Native drivers from Emulex or Qlogic perhaps? Assuming luns are presented from the chosen disk array then use cfgadm -c configure, job done. No old fashioned edits to sd.conf, just one command, doesn't matter which lun number, job done. Multipathing = edit a file, reboot and you've got free multipathing (mpxio) to almost all symetric or active/passive disk arrays. Some arrays don't adhere to T10 standards hence MPxIO may not like the arrays but most open storage arrays are fine.

ZFS from 10 onwards has it's place, probably first in data reliability from the checksumming perspective and goes onto sharing storage out the front end as well now (iscsi, comstar coming soon). VxVM/VxFS are add-ons but offer huge features but you pay for it.

As for debug of i/o 9 was a simple OS in that sense. 10 is way ahead of the game with Dtrace. Down to the thread style analysis of resource consumption from i/o to cpu to intterupts and so forth. We're not going to see that depth of probing elsewhere (correct me if you AIX guys have similar but I don't believe so) though I'm still scratching the surface with it and am still a bit edgy when probing live servers, always wondering if it's going to affect anything but good so far.

Virtualisation, well, sure. Sun went with hardware partitions and got into the logical partitions late in the game so ldoms are immature compared to IBM systems, I'd have to be mental to pretend otherwise. It's coming on, next releases slated for OS migrations like vmotion so seems to be catching up fast. Also looks like it's embedded for the future generations of servers coming through from Sun dev instead of FSC systems.

Zones can be considered ultra-reliable if clustered, some fast failover and elegant solutions can be made with N+1 server alongside Sun Cluster/VCS for failover zones. I'll be honest though and say I don't know if you can failover IBP or HP virtual partitions on power or superdome systems but I doubt it. Could be wrong.

Licencing, hmm, always a funny one. We're site wide for Oracle so it's not an issue, it's priced per user but the performance on the T2 boxes is pretty good. Digressing I never really understood how Oracle became the microsoft of the world and knocked out Sybase and so forth. I suspect Oracle aren't too helpfull to Sun on the per CPU licencing as they want to the core OS and cluster services layers so want to remove Sun from the equation, knock them from peoples minds as their previously favoured vendor.

The MySql aqquisition must have pissed them off as well.

Rightee, war and peace over.

Guys please stop commenting about products and features you have no experience with and frankly have no clue about them.

To Jesper and Bill: a ZFS clone and a disk clone are 2 different things!!!

cloning a ZFS filesystem is imediate (no file duplications involved)

cloning a ZFS filesystem is extremely space efficient, allowing you to have hundreds of clones on a single disk.

ZFS has checksums, compression, raid ....

ZFS IS FREE

Please also stop spreading the FUD about Oracle licensing on sun systems, you can manage your licensing costs using zones with resource caps, so you only need to pay for as much Oracle as you need.

I have experience with AIX and HPUX, and frankly my favorite is Solaris and Dtrace, ZFS, ZONES+ BRANDZ have no equivalent in HPUX and AIX.

And about th IBM power hardware, as with anything non X86 it will linger around, while x86 systems will take away market from them ....

I am no fan of X86 architecture, and nobody with a minimal hardware knowledge can be, however economics will always triumph ...

Dtrace has an equivalent on AIX 6.1. Its called Probevue.

Eseentially the same thing.

"I don't know if you can failover IBP or HP virtual partitions on power or superdome systems but I doubt it. Could be wrong."

Presume IBP is meant to be IBM ? Also presume that for virtual partitions, you mean the IBM equivalent of zones/containers which are called WPARs (workload partitions).

If my assumptions are correct, then I can confirm that WPAR's can be clustered using HACMP, along with normal virtualized and non-virtualized LPARs.

Running WPARs can also be dynamically migrated to a different hosting LPAR on a different physical system, which I dont think can be done yet with Zones/Containers. At least not without a shutdown, export/import etc.

Probevue the same as Dtrace? Please. Probevue is a relatively immature attempt at a clone of dtrace. Even IBM recommends against using Probevue in a production environment unless the performance hit is acceptable. Also, there are no where near as many providers in probevue as in dtrace, though I'm sure this will improve over time.

Probevue is not equivalent to dtrace... yet.

"my favorite is Solaris and Dtrace, ZFS, ZONES+ BRANDZ have no equivalent in HPUX and AIX.

"

erm...

On AIX we have Probevue, which isd Dtrace equivalent.

We haver WPARS, which is Zones equivalent. and better too ! :)

Ok, AIX does not have ZFS, but most of the features in ZFS already exist in JFS2, and have done for years. I will conceed the error checking is superior in ZFS currently, but I expect this will be addressed soon enough.

Brandz = Linux zones on Solaris global i beleive ? No AIX does not have this yet, but then i would say its a little premature to say Solaris offers this, as a quick google seems to suggest that this is still in beta ?? This topic was discussed at a recent IBM seminar though, and it is being looked at, as well as the running AIX 5.3 WPARS on AIX 6.1.

"Probevue is a relatively immature attempt at a clone of dtrace"

I think immature is a little harsh. I'll admit, i have not yet got my teeth into it in any depth, but it looks to be fairly comprehensive, if a little daunting.

http://publib.boulder.ibm.com/infocenter/systems/index.jsp?topic=/com.ibm.aix.genprogc/doc/genprogc/dynamic_tracing.htm

> Guys please stop commenting about products and features you have no experience with and

> frankly have no clue about them.

I am sorry, judging from the comments I am going to make and the comments made by other I'd rather say it's the other way around.

>To Jesper and Bill: a ZFS clone and a disk clone are 2 different things!!!

Yes it is. That is why I used the term functionality, and it was in the context of liveupgrade.

> cloning a ZFS filesystem is imediate (no file duplications involved)

> cloning a ZFS filesystem is extremely space efficient, allowing you to have hundreds of clones on a single disk.

> ZFS has checksums, compression, raid ....

> ZFS IS FREE

ZFS is great, ok perhaps a bit overhyped, but shouldn't we just stop at that, I've never said that it wasn't great product.

>Please also stop spreading the FUD about Oracle licensing on sun systems, you can manage your

>licensing costs using zones with resource caps, so you only need to pay for as much Oracle as you

>need.

That is not FUD. It is facts. if you deal with SUN sales people you would know as soon as Oracle and licensing pops up their eyebrows start to twitch. And no matter what you say then it doesn't change the fact that for a SPARC server it is 0.75 license per core. Sure you can limit the number of cores that Oracle can run on by partitioning your server into smaller bits, but it does not change the fact that you'll end up paying the same in license for a SPARC core as you do for a power6 core.

Now you mentioned that you could limit your license fee by partitioning up your server, and here is another advantage of the virtualization on POWER.

Let me try to explain, by using an example. You have 4 different DB servers. That have a total

maximum peak usage of 700K TPMC, at the same point in time. But as they peak at different points in time they have local peaks at 300K, 400K, 500k and 600K.

Now on a traditional partitioned server like a M8000 you would make one partition that could handle a peak of 300k, one of 400K etc. Hence the capacity you would allocate would be 1800 TPMC.

Now on a POWER server you would simply make a processor pool that could handle the maximum peak load, and a little bit more just to be on the safe side. Lets say 800K. On for example a p570 that would be 8 cores x 4.7/5.0Ghz CPU's. Now this means that you would never have to pay for more than 8 cores -> 6 licenses for Oracle.

Inside this pool you would then allocate resources to your virtual machines, you would then normally give the virtual machines more virtual CPU's than they needed and uncap them, hence allowing them to use more CPU power than they needed. For example 5 virtual cores,6 virtual cores,7 virtual cores and 8 virtual cores, at 100K TPMC per core. Hence the virtual machine that ran the 600K TPMC workload could actually peak up to 800K if need be.

Now on your traditional partitioned M8000 you would end up with something like this:

One partition with 9 cores.

One partition with 12 cores

One partition with 15 cores

and last

One partition with 18 cores

for a sum of 54 cores or a sum of 41 Licenses. If we assume 33K tpmc per core. (or 54 if we assume 25K tpmc per core which btw would mean that we had to use a M9000)

So to sum up 8 core p570 versus 54 Core M8000 and 6 licenses versus 41 licenses.

Now do you understand why I as an architect like IBM pSeries ?

> I have experience with AIX and HPUX, and frankly my favorite is Solaris and Dtrace, ZFS,

>ZONES+ BRANDZ have no equivalent in HPUX and AIX.

As stated otherwise you are wrong with regards to Dtrace.

>And about th IBM power hardware, as with anything non X86 it will linger around, while x86

>systems will take away market from them ....

Yes and the mainframe is dead, and so is Java and..............

> I am no fan of X86 architecture, and nobody with a minimal hardware knowledge can be,

>however economics will always triumph ...

No, that is where you are wrong. There are not many people that understand the economics of computer infrastructure cost. I read a study, think it was either Gartner or IDC, on the buying pattern of heads of IT. And the single most important factor was momentum in the marked place. Or as you would call it in other businesses 'What's in Fashion'. And you would be surprised on how often a Linux/UNIX on Sparc/Itanium/Power turns out to be cheaper than a x86 solution in TCO, if you have to make solutions that can live up to the same specs.

// Jesper