Grid ... Was: Is RAC DOA?
- From: Mogens Nørgaard <mln@xxxxxxxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Mon, 16 Aug 2004 22:56:20 +0200
Fundamentally, it's always a problem when a lot of bright people don't
understand immediately what a vendor means by a certain word.
It's either because it's so incredibly revolutionary in its entire
concept that the bright people, who usually understand most of what
comes out from this vendor, are left by the wayside. Or it's because
it's not what the vendor claims it is.
So let's say that the Grid idea is completely, utterly revolutionary,
and that most of us - for the first time in our Oracle-lives - don't
understand it. The amazing thing is that all Oracle-employees seem to
understand it (although they all have different ways of explaining what
it is and what it really consists of). This proves that Oracle's
employees are way, way smarter (not just smarter) than people not
employed by Oracle.
I'm not saying this is not the case. I'm just saying it's an amazing
fact in itself. I've argued before, that the general feeling inside
Oracle is that people outside of its firewalls are not, uhm, the
Einsteins of the World. In fact, I've noticed that the second people
walk out of Oracle's door to start a new job outside of Oracle, they
immediately become rather dim. Something happens in that door, Mladen!
The Grid is not a product, it's a concept. Well, that's nice. And I'm
sure saying that transportable tablespaces and RAC and various queueing
stuff allows you to construct a grid is true. But it becomes a sort of
circular argumentation. I'm pretty sure that just as we see the Oracle
RAC folks making fun of OPS today, a couple of years from now some
Oracle folks will make fun of the Grid they are telling people to
believe in today.
You can't buy something called a "grid" from Oracle. You can buy various
technologies from Oracle which allows you to put something together
which Oracle will then declare to be "a grid".
So the vision is to deliver computing resources like water or
electricity is being delivered today, where various producers can
supplement each other.
I do find fault with this comparison, which in my view is one of the
worst over-simplifications ever done by the S&M people of the World.
Imagine a world (Planet) where you're told that the current (sorry, joke
un-intended) electricity you're using is out-dated. You will need to go
to the newer version of electricity, which happens to be 342 Volts and
42 Amps - and the plugs in your wall will need to be "upgraded" to a new
model with seven different sized and shaped thingies to fit into each other.
Imagine a Planet where you're informed that Water version 1 is not
really the future. Instead, Water version 2 will be delivered beginning
of next Summer, and version 1 will no longer be supported. Version 2
will require a few changes in your house: First of all, you'd better get
used to the fact, that only hot water will be delivered. No cold water
anymore - but we suggest you buy an extra cold-water server for this
purpose in case you still require this out-dated service. Second, since
you will no longer need to have special handles for hot and cold water,
it is suggested to change all your appliances to new, more sturdy types
that can take all the hot water (which, by the way, comes with a much
higher pressure, so better check the state of your pipes).
IT is not like that. One day it might be, but I really cannot imagine
how it can be that. Requires much vodka.
Shifting resources is not a simple task, and many things have to be
considered. I'm sure Oracle is busy finding out how to address a lot of
these mysterious and wonderful possibilities. But it's stretching it a
byte and a bit declaring this to be a ready-to-buy thing.
When the explanation for something apparently very technical is filled
with S&M buzzwords there's a reason.
Mogens
MacGregor, Ian A. wrote:
> I'm a bit confused here. Hereogeneous means of different origin. I thought
> inside a cluster things had to be the same. They had to be homogeneous;
> i.e., of the same origin. I also thought the problem with Oracle is that you
> need to create a single large cluster instead of allowing you to have
> multiple clusters which can communicate with each other. So with the Orcle
> grid software, when you learn of hardware vendor going out of the machine
> business, and you can buy scores or hundreds of machines on the cheap. You
> have to pass.
>
>
> Ian MacGregor
> Stanford Linear Accelerator Center
> ian@xxxxxxxxxxxxxxxxx
>
> -----Original Message-----
> From: Daniel Fink [mailto:Daniel.Fink@xxxxxxx]
> Sent: Monday, August 16, 2004 11:46 AM
> To: oracle-l@xxxxxxxxxxxxx
> Subject: Re: Is RAC DOA?
>
> Extend that question into the realm of 'grid' computing. RAC requires a
> heterogeneous environment within the cluster. This means that a RAC cluster
> is not really grid friendly. It is only able to 'shift' processing
> requirements within itself and cannot take advantage of that underutilized
> server sitting next to it in the data center. I asked this question of an
> Oracle employee recently and the response was that RAC does indeed fit into
> Oracle's grid strategy, but there are not any plans to enable a homogenous
> RAC system. Am I missing something?
>
> Daniel
>
>
>
> Jesse, Rich wrote:
>
>>I'll add that until <buzzword>hetergeneous clusters</buzzword> are supported,
>>your availability is only good until your next patch (Oracle, OS, or 3rd
>>party). When you patch one, you most likely need to patch all at the same
>>time. In the case of Oracle patches, it's mandatory at least thru 9.2.
>>Rich
>>
>>-----Original Message-----
>>Sent: Monday, August 16, 2004 10:48 AM
>>Subject: RE: Is RAC DOA?
>>
>>
>>One slight quibble, regarding increased availability.
>>
>>[snip]
>>
>>mwf
>
>
> ----------------------------------------------------------------
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> ----------------------------------------------------------------
> To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx put
> 'unsubscribe' in the subject line.
> --
> Archives are at //www.freelists.org/archives/oracle-l/
> FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
> -----------------------------------------------------------------
> ----------------------------------------------------------------
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> ----------------------------------------------------------------
> To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
> put 'unsubscribe' in the subject line.
> --
> Archives are at //www.freelists.org/archives/oracle-l/
> FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
> -----------------------------------------------------------------
>
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
