Abandoning HPUX

[Mark Grant]

Hi all,

Today I find myself in an unusual situation. After spending eight years trying to persuade management to consider Linux, I find myself trying to persuade management to not use Linux.

We are close to capacity on all our systems with many more projects on the way. Management have asked me, yes you read that correctly, asked me, to evaluate the possibility of using Linux to address this capacity problem instead of new HPUX servers. The eventual aim would be to migrate everything off of HP and move to Linux, possibly on an intel platform.

My gut reaction surprised me. I felt this was not a good idea. I can run ten year old software on our current HP machines, I can not run software I bought on Linux three years ago.

The stability of HPUX, not just in PANICS per year but in it's interfaces is much higher than Linux in my view.

I would be grateful to hear your views on the subject and any experience of abandoning HPUX in favour of Linux.

Regards


Mark

[Pete Randall]

Mark,

April Fool!?!?


Pete

[Mark Grant]

:) Pete,

No, this is the real deal. Manager stood their bold as you like saying, "Maybe we should move to Linux". I don't know what magazine he's been reading recently but we suffer Management by magazine quite badly here.

[H.Merijn Brand (procura]

In that case, suggest M$Servers, and find yourself a new job

Enjoy, Have FUN! H.Merijn [ Who grew _very_ fond of HP-UX over the past decennium ]

[john korterman]

Hi Mark,
perhaps then your manager has read about how to make a good april's fool joke on your unix admins.....?

regards,
John K.

[Mark Grant]

Merijn, Maybe that's what he meant!! Maybe he thinks Linux is the latest version of Windows.

[Bart Paulusse]

Hi Mark,

well, a very important thing to consider is where to turn to for support.
If you run into some software issues for an HP-UX product, there is a professional organization that will solve your problem.
Who picks up your software issues when your running Linux on an intel platform. The "linux community" doesn't sound like a very "solid" platform for support to me.

regards,

Bart

[Karthik S S]

Mark,

If you are considering RH Linux:

I think your management is not aware about the new Redhat policy. Latest versions are not free anymore.

http://www.redhat.com/software/rhel/

-Karthik S S

[Hoefnix]

Mark,

If it's not a joke, you are still able to get support on Linux systems from HP (also on Intel).
I was on a Linux Roadshow of HP and they are promoting it very much.

So see it as a joke or as a challenge its up to you.

Regards,
Peter

[Mark Grant]

Karthik,

To be honest, if you've ever seen my posts in the Linux forum, you might guess that Red Hat would not be my first option anyway :)

However, the individual licensing cost is not the driving force here. Look at the cost of RAM, or another CPU for an rpXXXX machine. We need more RAM in two machines, a couple of FC cards in another machine, a CPU or two in another, maybe we should just throw in an intel box or three and run the new database on that. Plus, it appears that Linux is Oracle's development platform for 10.X if we can believe them.

There's a compelling and easy argument for cheap unix based hardware. It is difficult to frame a good argument for sticking with HP. I understand the "support" argument but my personal experience is that community based support (such as this forum) is ten times better and ten times faster than any vendor support (including Red Hat's) that I've used.

[Michael Tully]

The argument I always bring to the table is not always the cost of hardware, but the cost of change. What happens if you change platforms? Invariably software has to change, then procedures need to change. What about business practises. What is the cost of customisation? Training of not only user staff for new software, but also the personnel maintaining something with little or no experience. I'm not trying to sound negative, but realistic ....

(too bad it wasn't an april fools joke....)

[Hoefnix]

Mark,

You addressed Oracle on Linux.
Yes it's possible and runs fine. But I was speaking with a expirienced DBA about a big logistic application. He mentioned that there is still a 32 bit/64 bit issue. When using oracle on Linux it should still run in 32 bits mode he mentioned. Also the total amount GB mem that can be addressed using one single machine can be an issue.

HTH,

Peter

[Pete Randall]

Mark,

It it's not a joke and it's not about licensing costs but more about hardware costs, then I have to question the reliability aspect. I know my HP9000 boxes are rock solid and failures are handled first by redundancy designed into the box and second by trained hardware personnel. I rarely have downtime because of hardware issues. Can the same be said for Intel platforms - I don't think so.

There's also the issue of driver support for the latest Intel hardware. In my experience the latest and greatest generally hasn't had a Linux driver developed for it yet and you end up spending countless hours fighting to get Linux installed, but you know about that.

My main objection would be the reliability and service aspect.


Pete

[Shaikh Imran]

Hi,
This is a very sad news that someone is trying to migrate from HP to Linux just for Cost sake.
My personal feelings are same as yours for HP.
Linux,what i think is not a good choice if
1)we try to use 2GB Fiber channel cards-- Intel is Intel after all..
2)I agree the support point what BART says
which i have also faced earlier.
3)What if you want to Use a Storage SAN or NAS Boxes
4)What if you want to USe ULTRIUM Libraries..
You will have to struggle a lost with Linux
if you want to use sophisticated equipments.
5) Why want to Risk the Company's Data which is Worth 100 times the Hardware& Software Cost...

Anyways it will be good if your management drops the idea and look for the upgrades of
the current hardware....

Hope for the Best!
Regards

[Fred Ruffet]

Problems about linux is that administration is oftenly done in an other way than on Unixes. You are not supposed to change your kernel every day... By choosing a distribution like slackware (which aims at security and stability), you can have something that is really working on a production environement.
On Slackware 9.0, Oracle 9i installs whithout any patch, and works great.

[Elmar P. Kolkman · Accepted solution]

Mark,

On hardware level, Intel based hardware can be cheaper. And the support too. But mind that a lot of HP hardware has a lifespan well over what any intel box has ever had (anyone still using P-II boxes in production? And how many are still using K or D class systems?)

As for the stability of the hardware: redundant power supplies etc are no issue for intell hardware anymore. For instance, look at the poweredge series of Dell.

So, hardware is not a real issue (though most is still 32-bit, which might be an issue). So we have a look at the software.

As for the OS, support is not an issue if you have a short lifespan in mind. But the support of an Linux version stops after the next two releases, normally. For instance, 3 years ago the RH version was 6 or 7.0 while HP-UX version was 11.0. Try to get support for the RH version ;-)

But this issue is partially solved: RH started Advanced Server and I'm sure other distributions have solutions like this.

Leaves the software you need to run under Linux. Peter already pointed out a problem with oracle. Other software is available, but not is well tested on Linux as on HP-UX. And before migrating, make sure it is really completely available for linux. For instance, VitalQIP only has a client, no server software for linux. Legato Networker had the same problem till some time ago. Other software might have the same problem.

As for the panics-a-year status: that totally depends on what you do. We have systems here running more then a 1000 days without a reboot. I've seen linux systems with the same uptime, but those are sporadic...

In the end you have to see for yourself, but I would not replace HP-UX yet. But starting to test it or implement new applications/versions on linux might be a way to go.
Good luck with educating the management.

Elmar (who hopes to be back in the chatroom soon)

[Elmar P. Kolkman]

A reason to go to linux: Management likes good figures. And running unix on boxes with 3Ghz processors looks a lot more impressive then the 750Mhz of a PA-Risc ;-)

[Mark Grant]

Guys,

There are some good answers here so if you don't mind I'll defer the point assignment for a little while to keep a bunny off the thread.

We do actually run Linux here. We have about twenty Linux servers that are used to consolidate about sixty windows machines (we run our Windows servers as virtual machines in vmware on Linux) so the support/training issue is not so strong for us. Plus we are in the process of off-shoring everything too so we may not even need to admin the boxes ourselves. Management just do not see that there is a cost in adding to our (now nearly 100) databases.

I'd be particularly interested to hear in more experience on the Linux and SAN issue's (we have ours SAN attached) and the Oracle 32 bit issue (is this true for 10.x)

Fred,

I'm with you on that one, my desktop is Slackware and it's my distro of choice :)

[Hazem Mahmoud_3]

We just recently had a similar discussion here in the IS department. We currently run HP-UX for our core system and we have a plethora of Linux and Windows servers for our other 3rd-party apps.
We intend on putting as much of our 3rd-party apps on HP-UX as possible. Eventually we would like to reduce the number of Linux and Windows servers as much as possible.
The major reason was due to support and history. HP-UX has a much longer history than Linux does, therefore it is more stable. We also concluded that to have a company like HP to back us up in a disaster is much more comforting than having some Linux support company.
Not to say that Linux isn't the right solution, but I think it just needs more time and stability. I think that the move by other companies like HP, Dell, IBM, and Novell to adopt linux in their systems that they sell is proof that Linux has a future.

-Hazem

[Umapathy S]

Hi Mark,

IMHO, Linux is not mature yet for the Enterprise market. Even hundreds of Intel PCs clustered together wont replace a rock solid hp PARISC boxes. But they are trying to break that with Itanium chips. May be Linux on Itanium be better but that is yet to be tested and proved in market.

My plan is to introduce linux into a small part of the infrastructure (like a test bed where a small amount of the total requests go into them) and increase the tempo slowly. It will help to consolidate in the long run.

cheers
Umapathy

[Richard Darling]

Mark,
Were you listening to the Howard Stern show this morning, and thought you'd pull one here, too?

If HPUX has been running well for ten years and you have problems with 3 yr old apps on Linux...what else do you need to tell your mgt?

Good Luck.
Richard

[Fred Ruffet]

For the SAN part :
Sorry to tell this on a HP forum, but I did set up a linux (RH AS 2.1) cluster on IBM Xseries runing Oracle RAC on a FastT500 SAN. It took time to have every component be supported. IBM did only support 2.4.9 kernel whereas Oracle wanted 2.4.13... As long as IBM really wants to support linux, they made real efforts, and we gained support for all (RH part supported by IBM).
It was a really good platform, and supported a huge number of connections without any problem...
I suppose HP would do the same for their Intel server. I would like to see it with a linux on ia64...

[Geoff Wild]

Here's a little post I found about using Linux (RedHat) in the Enterprise:

--------------------------------------

What Is an Enterprise Distribution?
High customer expectations have been set by the proprietary UNIX operating systems, and customers planning to migrate from UNIX to Linux do not want to adopt technology that cannot deliver the same level of robustness, quality, support and compatibility. Business users demand stability and reliability. In some cases this means bleeding-edge technology is not appropriate for inclusion in a product. Users also want the ability to run across a wide range of architectures and hardware components, thereby realizing the Linux goals of avoiding proprietary vendor lock-in. Support needs to be in the form of ongoing maintenance for several years, including security and bug fixes, as well as incremental hardware support and valuable, but not destabilizing, feature enhancements.

Requirement Gathering
The ball gets rolling on a new release by identifying the targeted feature set. Next to strong opinions, the second most plentiful commodity at Red Hat is feature requests. They come from all directions, such as insatiable independent hardware/software vendors, demanding customers, both large and small, and Red Hat's worldwide sales and service support organizations. Additionally, as many ideas are generated within Red Hat engineering, ranging from performance and usability enhancements to marketing proposals on how to organize the product set.

Obviously, we don't have infinite developer resources, so the major challenge is to choose the best of these ideas. Another key feature acceptance criteria is conformance with upstream Linux direction, which is necessary for compatibility and to remain true to the spirit of the kernel.org tree governed by Linus Torvalds and containing contributions from around the world.

Here's a few examples of the challenging scenarios we face in requirements gathering:

We ask each of our partners to submit a reasonably sized top-ten list. One partner's requirements came in the form of a binder that was two inches thick. This became affectionately known as the bible. My first exposure to the bible came when it was heaved onto my desk with a resounding thud. When I saw that, I swear that my heart went thud, too.

The more mathematically inclined partners do internalize the concept of a top-ten list. However, most ended up using a tactic that those of you familiar with TCP/IP networking should recognize the sliding window protocol. The way it works is as soon as any of your features have been accepted, those pop off the top of the stack, freeing up space for features 11, 12 and 13 to all of a sudden become cataclysmic issues.

The following is a representative example feature list from a hardware vendor:

Support more than 32GB of memory on x86.

Support more than eight CPUs.

Support for our new XYZ100 series computers. (This is a thinly veiled multiple feature request; behind it is a series of required device drivers, PCI IDs and installer hooks.)

Updated I/O adapter driver. (This ultimately turns into differences of opinions regarding to which newer version this refers).

Integrated support for the vendor's proprietary baseboard management software. (A perennial list item, which consistently gets rejected to the amazement of the requester.)

Compiler optimizations to match the latest chipsets.

USB support for our CD-ROM drive. (Needed because the drive is brain-dead and doesn't conform to the spec - of course that subtlety is absent in the initial feature request.)

Support for more than 128 disks.

Then, there's typically the following implied requirements:

All of these feature requests apply to multiple architectures, including x86, AMD64 and Itanium 2.

Oh, by the way, we also want this feature backported to the prior Enterprise Linux v. 2.1 release.

Countless external requests ask for either proprietary additions or hooks. In the open-source tradition, this is something we consistently have to refuse.

--------------------------------------

Bottom line is - if you can do what needs to be done on Linux - then why not? If no - then stick with HP-UX. For example, http://www.sap.com/linux/ states you can run mySAP Business Suite on Linux...will it be as stable? I don't know. Cost is a big factor - and I can see why your managers are looking to you to come up with a Linux solution. Why spend a $200,000 on hardware (RP7410) when you can accomplish the same on $33,000 (ProLiant ML570)?

You can still go with HP on some things - even MC/SG:
http://h18004.www1.hp.com/solutions/enterprise/highavailability/linux/serviceguard/index.html
Only $1,167 per node.

Rgds...Geoff

[G. Vrijhoeven]

Hi Mark,

Just out of curiosity, what are the arguments for your management for this request. Is it cost driven ( i can not imagine 64 Bits DB's so Ithanium hardware on SAN disks, with a now OS breed, migration down time, new work instructions, re-education for admins is cheaper). support driven, bad experience in support or support rates or is it time for a new support agreement and is the management looking for options, just to have an alternative.

HTH,

Gideon

[Mark Grant]

Gideon,

It is really a kneejerk reaction to having no capacity planning and no understanding that you can't keeep throwing databases at machines without increasing your infrastructure costs.

In a nutshell, they don't like the hardware costs of buying new rpXXX's, RAM and other bits of hardware.