Re: Oracle internal flaws?

From: Niall Litchfield <niall.litchfield_at_gmail.com>
Date: Tue, 15 Mar 2011 16:13:29 +0000
Message-ID: <AANLkTi=VifvaSjO_104_Eb+PSoRUEWj=2yzciL5LPY9r_at_mail.gmail.com>



I've commented on this post
http://stackoverflow.com/questions/5307590/cpu-usage-of-oracle-installed-database-machine/5314571#5314571 -
so far as I can understand what he means every single one of his comments about Oracle's architecture has been wrong for at least a decade and mostly about 2. And of course spouting nonsense about "architecture" when you don't know or measure what is happening is somewhat bone headed.

On Tue, Mar 15, 2011 at 3:25 PM, Goulet, Richard <Richard.Goulet_at_parexel.com
> wrote:

> Mark,
>
> After looking at this individuals page I have serious doubts that he's
> anything but a fake and possibly a looser in the IT world:
>
>
> 34 years IT (started in 1977 during the punch card days) , 32 years
> databases (started in 79? That's pretty old stuff) , 19 years banking and
> finance, 19 years Sybase, Partner for 13 years but gave it up this year with
> the SAP changes.(why did SAP change the world that badly?) Worked for
> DBMS Vendors (none specified) , maintained DBMS codeline (what code) ,
> wrote one from scratch for them (really, I've a bridge in NY to sell too);
> consultant since then (consultants are either good, or just plaques on the
> wall. I think this is a wall plaque).
>
> Intolerant of contradictions in the same cranial space.
> *Proper Preparation Prevents Poor Performance*
> If it's worth doing, it's worth doing Right
>
> Extreme performance in every technical endeavour. The second greatest buzz
> is increasing the speed of a customer's server by an order of magnitude. Or
> two.
>
> Wheels are there for improving, not re-inventing. Standards
>
> The rest sounds like just plain crankiness/intolerance. In general he/she,
> we don't know, sounds like an old foggy, like me, who hasn't changed with
> the times. Bet he still codes in Assembler and believes in host based
> computing with terminals. Probably likes punch cards as well. BYW, a look
> at his listed web site says that he's a Sybase only kind of guy and at last
> look Sybase was loosing a good pile of market penetration to Oracle, DB2,
> and SQL*Server. They're also not pushing their db as much as they use to
> and more into the app front end. Sounds like a real pile of sour grapes and
> someone who really doesn't know what he's talking about. I would not like
> to get into a Sybase discussion with him as he probably does know more about
> it than most people on this list, but that's where it ends.
>
> Now for the fools points:
>
> Oracle does not have a true server architecture (others have it). Rather
> than performing classic server tasks, such as multi-threading, caching of
> data pages, parallel processing (split a query across many devices) etc.
> within itself, it uses the o/s to do all that. That means for each user
> process (PL/SQL connection) there is one UNIX process; 1000 users means 1000
> UNIX processes, all competing for the same resources.
>
> Obviously he has not heard of MultiThreaded server, parallel processing,
> and the SGA. Now while a lot of this was true in very old versions of
> Oracle (v3 was all dedicated, all PGA, parallel didn't exist for anything
> including the OS, and Oracle was a single process on the OS. We also had
> before and after image files, not memory areas.) it is no longer the case.
> Today we have some 10 separate background processes that coordinate with
> each other. We also have a job queue and a job scheduler that handle jobs
> both inside and outside the database. BIF and AIF files have turned into
> REDO segments in their own tablespace, and the db cache is much more
> efficient and flexible to boot. Needs to update his knowledge, badly.
>
>
>
> Especially noteworthy, because it uses file system files (not raw
> partitions), and the "caching" is outside, it relies heavily on (and is very
> sensitive to) the file system cache that you have set up. likewise, Oracle
> needs a massive amount of memory for these processes.*”*
>
> Well I'll give on the memory issue, Oracle has always been a memory hog and
> it's getting worse all the time, course memory is a whole lot cheaper today
> then it was in the 80's. Back in the good old days you had to add an 2MB
> expanded memory card and drivers to DOS pc's to run Oracle. As for raw
> partitions, they are "faster" than a cooked file system but less flexible
> which is why Oracle came out with ASM. Also many sysadmins I've run into
> don't like raw devices either. You can't easily see, so I'm told, what's
> going on down in that device as in how much is in use and how much is free.
> So use ASM and you can have the performance of raw with the flexibility of
> cooked.
>
> One bad points of the internet is that anyone with no knowledge can
> promote themselves as know it alls. I learn something new every day from
> this forum as well as others and the manuals that we love to diss as well as
> practical experience. I don't profess to "know it all" and I encourage
> anyone to point out the flaws.
>
> Dick Goulet
> Senior Oracle DBA/NA Team Leader
>

-- 
Niall Litchfield
Oracle DBA
http://www.orawin.info

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Mar 15 2011 - 11:13:29 CDT

Original text of this message