I am starting a Proof of Concept project to demonstrate benefits of
migrating our Oracle databases from Windows to Linux. This is driven
by potential performance improvements.

I am trying to decide on a particular flavour of Linux.

Since this is an Enterprise wide application, our choice will be
limited to the Oracle supported platforms, i.e.:

Redhat Enterprise AS (2.1, or 3)
Redhat Enterprise ES (2.1, or 3)
Suse Linux Enterprise Server
United Linux 1

Does anyone have any experience/info as to whether any one of the
above is 'better' from a support, stability and performance
perspective. These are my 3 major concerns.

The HW will be IBM xSeries servers connected to a Hitachi S.A.N.

There is also the possibility of trialing 64-bit flavours of
Linux/Oracle on Itanium.

My instinct is to go with Redhat because I have had more sys admin
experience with this flavour.

Any feedback welcome

Matt

> Redhat Enterprise AS (2.1, or 3)
> Redhat Enterprise ES (2.1, or 3)
> Suse Linux Enterprise Server
> United Linux 1

whitebox for dev
RHEL 3 for enterprise

Matt wrote:

> I am starting a Proof of Concept project to demonstrate benefits of
> migrating our Oracle databases from Windows to Linux. This is driven
> by potential performance improvements.
>
> I am trying to decide on a particular flavour of Linux.
>
> Since this is an Enterprise wide application, our choice will be
> limited to the Oracle supported platforms, i.e.:
>
> Redhat Enterprise AS (2.1, or 3)
> Redhat Enterprise ES (2.1, or 3)
> Suse Linux Enterprise Server
> United Linux 1
>

Solid:

- RedHat ES 3 for machines less than or equal to 4 CPU
- SuSE SLStdS8 for machines less than or equal to 4 CPU
(Suse Linux Standard System is much less $ than SLES)

- RH AS 3 for enterprise > 4 CPU
- SLES8 for enterprise > 4 CPU

- SLES9 for anything that Oracle has certified! See Oracle's Certify site

- WhiteBox Linux for development as it mimics RH3AS.

Avoid like the plague any of the freebie 'public' distros like Debian,
Fedora, SuSE Pro, Mandrake, and so on. While they are extremely good and
beneficial, the amount of tinkering needed to get Oracle stable, and the
rapid release cycle, make these unsuitable for you environment. I run
several of the freebies, but NOT under Oracle! Reason - anything other
than the basic database will likely give hicups, occasionally at random.


Preferred by me:

I personally am getting very happy with SLES9 for Oracle now that it's
certified. Since it's late to the game, not all Oracle products are
certified on SLES9, but I think the certifications are being done as
quickly and carefuly as possible.

There are a few things to watch for with SLES9, but they are getting fixed
rapidly. I personally like SuSE's support better than RedHat - ymmv as it
depends _entirely_ on the individual on the other end of the line.


Comment:

Many people mistakenly believe that United Linux was a distribution. It was
a _specification_ with a core kernel and minimum distribution requirements
to be used as the basis for distributions. SLES8 is SuSE's implementation
of United Linux 1. Now that UL is basically defunct, the consortium
members are independant - SLES 9 is a wonderful distro and perfect example
of where UL should have been going.

/Hans

HansF wrote:

Won't disagree with anything you wrote except...

> - WhiteBox Linux for development as it mimics RH3AS.

It doesn't "mimic" RHAS3, it *is* RHAS3. Exact same source. Just RH
proprietary logos removed. "Mimic" makes it sound like it's an emulation,
or an approximation, rather than what it truly is: a clone.

A minor quibble, I agree.

Regards
HJR

Howard J. Rogers wrote:

> HansF wrote:
>
> Won't disagree with anything you wrote except...
>
>> - WhiteBox Linux for development as it mimics RH3AS.
>
> It doesn't "mimic" RHAS3, it *is* RHAS3. Exact same source. Just RH
> proprietary logos removed. "Mimic" makes it sound like it's an emulation,
> or an approximation, rather than what it truly is: a clone.
>
> A minor quibble, I agree.
>
> Regards
> HJR

Counter-quibble .... even 'they' say it is _derived_ from RHAS3. From the
http://www.whiteboxlinux.org/

"This product is derived from the Free/Open Source Software made available
by Red Hat, Inc but IS NOT produced, maintained or supported by Red Hat.
Specifically, this product is forked from the source code for Red Hat's
_Red Hat Enterprise Linux 3_ product under the terms and conditions of it's
EULA."

The only reason for my 'mimic' statement is that WhiteBox is not obligated
to (nor has the resources for) keeping up with RH's service packs. In
addition, they are creating and providing their own patches which may not
be retrofit to RHAS. Thus it is a true fork, not an exact duplicate.
Anyone who does not recognize that may find themselves in a bind down the
road, even though they are OK for now.

WhiteBox is now in exactly the same position as Mandrake was several years
ago, when *that* was forked from RedHat.

/Hans

HansF wrote:

> Howard J. Rogers wrote:
>
>> HansF wrote:
>>
>> Won't disagree with anything you wrote except...
>>
>>> - WhiteBox Linux for development as it mimics RH3AS.
>>
>> It doesn't "mimic" RHAS3, it *is* RHAS3. Exact same source. Just RH
>> proprietary logos removed. "Mimic" makes it sound like it's an emulation,
>> or an approximation, rather than what it truly is: a clone.
>>
>> A minor quibble, I agree.
>>
>> Regards
>> HJR
>
> Counter-quibble .... even 'they' say it is _derived_ from RHAS3. From the
> http://www.whiteboxlinux.org/
>
> "This product is derived from the Free/Open Source Software made available
> by Red Hat, Inc but IS NOT produced, maintained or supported by Red Hat.
> Specifically, this product is forked from the source code for Red Hat's
> _Red Hat Enterprise Linux 3_ product under the terms and conditions of
> it's EULA."
>
> The only reason for my 'mimic' statement is that WhiteBox is not obligated
> to (nor has the resources for) keeping up with RH's service packs. In
> addition, they are creating and providing their own patches which may not
> be retrofit to RHAS. Thus it is a true fork, not an exact duplicate.
> Anyone who does not recognize that may find themselves in a bind down the
> road, even though they are OK for now.
>
> WhiteBox is now in exactly the same position as Mandrake was several years
> ago, when *that* was forked from RedHat.
>
> /Hans


OK then: let's agree to use the technical term 'fork', though it's a word
that has little meaning (in this context) outside the rareified atmosphere
of Linux distros. At least within those circles its meaning is unambiguous,
and is therefore correct.

We could continue to discuss whether 'clone' or 'mimic' is a more accurate
-closer- description for the general public of a distro fork, and I presume
we would continue to disagree.

The point for me is that White Box is built from the RHAS3 sources, and its
patches are similarly built from Red Hat's own Errata srpms... meaning,
incidentally, that they do NOT "create and provide their OWN patches which
may not retrofit to RHAS", not at an rate in the sense of knocking up their
own code.

That is decidedly NOT what Mandrake does or did several years ago, it seems
to me. White Box's goal (however they choose to dress it up) is to be a
free RHAS3. Mandrake's was to be a better distro.

If they took RHAS source to start with; and if they patch it in the future
with RHAS source; and if they write no source of their own; then it seems
to me it walks, swims and quacks and therefore is RHAS3, not something
merely pretending to be it or seeking to approximate it (the usual
implications of the word "mimic").

Regards
HJR

Howard J. Rogers wrote:

> HansF wrote:
>
>> Howard J. Rogers wrote:
>>
>>> HansF wrote:
>>>
>>> Won't disagree with anything you wrote except...
>>>
>>>> - WhiteBox Linux for development as it mimics RH3AS.
>>>
>>> It doesn't "mimic" RHAS3, it *is* RHAS3. Exact same source. Just RH
>>> proprietary logos removed. "Mimic" makes it sound like it's an
>>> emulation, or an approximation, rather than what it truly is: a clone.
>>>
>>> A minor quibble, I agree.
>>>
>>> Regards
>>> HJR
>>
>> Counter-quibble .... even 'they' say it is _derived_ from RHAS3. From the
>> http://www.whiteboxlinux.org/
>>
>> "This product is derived from the Free/Open Source Software made
>> available by Red Hat, Inc but IS NOT produced, maintained or supported by
>> Red Hat. Specifically, this product is forked from the source code for
>> Red Hat's _Red Hat Enterprise Linux 3_ product under the terms and
>> conditions of it's EULA."
>>
>> The only reason for my 'mimic' statement is that WhiteBox is not
>> obligated
>> to (nor has the resources for) keeping up with RH's service packs. In
>> addition, they are creating and providing their own patches which may not
>> be retrofit to RHAS. Thus it is a true fork, not an exact duplicate.
>> Anyone who does not recognize that may find themselves in a bind down the
>> road, even though they are OK for now.
>>
>> WhiteBox is now in exactly the same position as Mandrake was several
>> years ago, when *that* was forked from RedHat.
>>
>> /Hans
>
>
> OK then: let's agree to use the technical term 'fork', though it's a word
> that has little meaning (in this context) outside the rareified atmosphere
> of Linux distros. At least within those circles its meaning is
> unambiguous, and is therefore correct.
>
> We could continue to discuss whether 'clone' or 'mimic' is a more accurate
> -closer- description for the general public of a distro fork, and I
> presume we would continue to disagree.
>
> The point for me is that White Box is built from the RHAS3 sources, and
> its patches are similarly built from Red Hat's own Errata srpms...
> meaning, incidentally, that they do NOT "create and provide their OWN
> patches which may not retrofit to RHAS", not at an rate in the sense of
> knocking up their own code.
>
> That is decidedly NOT what Mandrake does or did several years ago, it
> seems to me. White Box's goal (however they choose to dress it up) is to
> be a free RHAS3. Mandrake's was to be a better distro.
>
> If they took RHAS source to start with; and if they patch it in the future
> with RHAS source; and if they write no source of their own; then it seems
> to me it walks, swims and quacks and therefore is RHAS3, not something
> merely pretending to be it or seeking to approximate it (the usual
> implications of the word "mimic").
>
> Regards
> HJR

I will agree to the term fork. I was using the term mimic in the sense of
'acting the same, looking different' and they are certainly changing the
logos to look different.

There are also a few minor things different - "How is WhiteBox Linux
updated? ... Obviously RHN isn't available as an option, so yum has been
brought in from Fedora along with a version of up2date that can work with
yum"

There is also a concern about whether there is any software to which RedHat
retains a copyright. It happened with SuSE's YAST, although I believe YAST
has now been released under GPL. Once bitten ...

As to the statement about not creating their own patches ... from their site
http://www.whiteboxlinux.org, I quote

"What can you do to help?

* Download and install White Box. Poke it in unexpected places and see
what breaks. Report problems so they can get fixed. As with all Open Source
efforts, bug reports are good, patches are better still. "

The last 4 words are of interest. My interpretation is that they will
permit divergence of the software, albeit while trying to maintain their
goal of being compatible with _RHEL3_ until October 2008. (We have both
incorrectly stated RHAS - they aim for RHEL compatibility.)

Amusingly, they do also state the will maintain compatibility with a union
of RH -AS, -ES, -WS, -PWS and so on. This implies some compromises, even
if it aims to be 100% binary compatible with RHEL.

I will continue to use WhiteBox as the alternative to RH 'commercial'
versions for development, test and training environments because I am
comfortable that it is close enough. However, I will also continue urge
potential users to use it with eyes wide open as there are no guarantees of
compatibility (it's a goal, not a guarantee). There may also be timing
issues in releases of errata srpms that could affect users in production.

I am sure you could (probably will) point out other flaws in my statements.
The last word is yours. <g>

/Hans

HansF wrote:

[snip]

> I will agree to the term fork. I was using the term mimic in the sense of
> 'acting the same, looking different' and they are certainly changing the
> logos to look different.

My belief so far is that a couple of logos is all the difference. But
whatever...

> There are also a few minor things different - "How is WhiteBox Linux
> updated? ... Obviously RHN isn't available as an option, so yum has been
> brought in from Fedora along with a version of up2date that can work with
> yum"

But the first thing I do with my install is install XINE. Same deal, is it
not? It's now not 'virgin' RH. But the kernel/core/whatever still is.

> There is also a concern about whether there is any software to which
> RedHat
> retains a copyright. It happened with SuSE's YAST, although I believe
> YAST
> has now been released under GPL. Once bitten ...

Fair enough...

> As to the statement about not creating their own patches ... from their
> site http://www.whiteboxlinux.org, I quote
>
> "What can you do to help?
>
> * Download and install White Box. Poke it in unexpected places and see
> what breaks. Report problems so they can get fixed. As with all Open
> Source efforts, bug reports are good, patches are better still. "

Well, we read that differently. I read that simply as encouragement to
contribute to general open source efforts, not a statement that they will
release their own patches. They're a public library in the middle of
nowhere... I doubt they will be releasing anything very much of their own
volition any time soon!

> The last 4 words are of interest. My interpretation is that they will
> permit divergence of the software, albeit while trying to maintain their
> goal of being compatible with _RHEL3_ until October 2008. (We have both
> incorrectly stated RHAS - they aim for RHEL compatibility.)

I get confused. I'm not entirely sure what the difference is anymore. Care
to shed enlightenment to me and everyone else? RHAS21, wasn't it? "Advanced
Server 2.1". The "3.0" Red Hat thing was also "Advanced Server", I thought.

> Amusingly, they do also state the will maintain compatibility with a union
> of RH -AS, -ES, -WS, -PWS and so on. This implies some compromises, even
> if it aims to be 100% binary compatible with RHEL.

I give up at those concentrations of alphabet soup!

> I will continue to use WhiteBox as the alternative to RH 'commercial'
> versions for development, test and training environments because I am
> comfortable that it is close enough. However, I will also continue urge
> potential users to use it with eyes wide open as there are no guarantees
> of
> compatibility (it's a goal, not a guarantee).

I 1000% agree with that. Close enough for me to say it's the same, because I
like giving the essential gist, even if it's not the microscopic technical
truth. But short of parting with vast wads of cash, you are definitely on
your own as far as the future is concerned.

> There may also be timing
> issues in releases of errata srpms that could affect users in production.

Lord. I hope no-one really uses this in production!

> I am sure you could (probably will) point out other flaws in my
> statements. The last word is yours. <g>

The game here is to help enlighten everyone, not to point out flaws. I hope
the last word will actually be yours, because I for one don't understand
the subtleties between an AS and and ES...

Regards
HJR

Howard J. Rogers wrote:

> The game here is to help enlighten everyone, not to point out flaws. I
> hope the last word will actually be yours, because I for one don't
> understand the subtleties between an AS and and ES...

The difference between RH 2.1 and 3 is similar to Oracle 8 vs 9i. Newer
technology, etc. RedHat tries to release on an 12-18 month schedule with
patches to keep the basic product stable but not change the design concept.

The difference between EL and AS attempts to be similar to the difference
between Oracle Standard and Enterprise, but is closer to the difference
between Standard-One and Standard. From
http://www.redhat.com/software/rhel/comparison/

AS can be used on systems up to 16 CPUs.
EL can be used on systems up to 2 CPUs.

At one time, I believe there were also different admin modules. And of
course, there is a huge difference in available support prices. On the
comparison chart, you can drill into more detail, but the details all seem
a lot closer between EL and AS than they did back in 2.1

BTW - similar thing for SuSE ... SuSE Linux _Standard_ Edition for 1-4 CPU
machines and SuSE Linux _Enterprise_ Edition (SLES) for up to 16 CPU.
(Although they to have been renaming)

/Hans

mccmx@hotmail.com (Matt) wrote in message news:<cfee5bcf.0411120244.7b19587b@posting.google. com>...
> I am starting a Proof of Concept project to demonstrate benefits of
> migrating our Oracle databases from Windows to Linux. This is driven
> by potential performance improvements.
>
> I am trying to decide on a particular flavour of Linux.
>
> Since this is an Enterprise wide application, our choice will be
> limited to the Oracle supported platforms, i.e.:
>
> Redhat Enterprise AS (2.1, or 3)
> Redhat Enterprise ES (2.1, or 3)
> Suse Linux Enterprise Server
> United Linux 1
>
> Does anyone have any experience/info as to whether any one of the
> above is 'better' from a support, stability and performance
> perspective. These are my 3 major concerns.
>
> The HW will be IBM xSeries servers connected to a Hitachi S.A.N.
>
> There is also the possibility of trialing 64-bit flavours of
> Linux/Oracle on Itanium.
>
> My instinct is to go with Redhat because I have had more sys admin
> experience with this flavour.
>
> Any feedback welcome
>
> Matt

Matt,

attempt to obtain a 5 seat RHEL development license.
It will cover you for WS, ES and AS.
This way your server and your laptop can run the same kernel, so you
can reproduce your test cases more cleanly.

yes, its not as low of a cost as whitebox linux, but its very
affordable.
You'll have an easier time with the suits justifying that you are
testing on a supported, commercially available platform that can be
shipped pre-installed from an OEM such as Dell, HP, IBM and alot of
other Linux-friendly smaller manufacturers that attend expositions
like LinuxWorldExpo.

hth.

-bdbafh