Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Mailing Lists -> Oracle-L -> Re: Fav. Urban Legend...Mem vs Disk
I can testify to that. Connor might not be 100% Danish, but he's sort of
OK anyway.
My personal story regarding the wait stuff (the Wait Interface as I usually call it) dates back to my days in Oracle many moons ago. On a big, internal list there, Kyle Hailey (who now works for Quest) talked about this new way of finding bottlenecks. Next, I stumbled upon Anjo's YAPP paper and that was it. I nearly fell off the ferry I was on while reading it. Since then, it has never ceased to surprise me that people are still reading and beleiving the books (like the ultimate books from the ultimate people, and all the others) where ratio after ratio after ratio is discussed. Where pure guesswork and magic are described as being scientific. Yuk.
Cary Millsap, I think, first coined the phrase "Checklist tuning". That's exactly what most tuners, DBA's and other nice people are wasting their time doing, instead of following the simple rules of looking at where the time goes. Without knowing where the time goes, how can anyone conclude anything?
But man, it gives us lots of work as long as people are reading those books. Perhaps the OakTable should write a book full of bad advise (why shouldn't we if anybody else can get away with it) and then lean back and wait for readers to call us :-))).
Mogens
Connor McDonald wrote:
>I think you may have missed my sarcasm - I've been on
>the anti-cache hit ratio bandwagon for a long time...
>
>Cheers
>Connor
>
> --- DENNIS WILLIAMS <DWILLIAMS_at_LIFETOUCH.COM> wrote:
>
>>Connor - Cary Millsap presented the results of 10
>>trace files in a Hotsos
>>seminar I attended. The ratio ranged from a high of
>>108.57 down to a low of
>>0.79. The point is that the ratio is nowhere near
>>the oft-quoted 10,000.
>>This means that logical I/Os are not insignificant.
>>Even if physical I/O
>>were eliminated (all blocks cached, 100% cache hit
>>ratio), response time
>>would not drop to zero. This is why the emphasis in
>>tuning is shifting from
>>simple ratios to examining wait times. If the most
>>significant wait time is
>>physical I/O, then changing that will improve
>>overall performance. But if
>>the most significant wait time lies in another area,
>>then you may make
>>significant improvements in physical I/O and still
>>not improve overall
>>performance. I certainly wouldn't claim to be an
>>Oracle tuning expert, but I
>>believe that the new ideas on tuning that are
>>emerging provide a significant
>>step forward in making Oracle tuning more of a
>>logical process than a
>>collection of rules of thumb.
>>Dennis Williams
>>DBA
>>Lifetouch, Inc.
>>dwilliams_at_lifetouch.com
>>
>>
>>-----Original Message-----
>>Sent: Wednesday, March 20, 2002 3:49 AM
>>To: Multiple recipients of list ORACLE-L
>>
>>
>>Some rudimentary testing on a laptop here (500Mhz,
>>512M RAM, typical single disk)
>>
>>a) visiting a single block via 4,000,000 logical
>>IO's
>>got me approx 35000 gets/sec
>>
>>b) repeated full table scans similar system got me
>>approx 350 phys reads/sec
>>
>>After this extensive, thorough and exhaustive
>>exercise, I can definitely say that memory access
>>versus disk access (as it pertains to Oracle) is 100
>>times faster on this machine in single user mode
>>
>>I think we can generalise this to be the rule for
>>all
>>servers under all conditions :-)
>>
>>Connor
>>
>> --- "Freeman, Robert " <Robert_Freeman_at_csx.com>
>>wrote: > I've heard the disk vs. memory arguments
>>before, but
>>
>>>never have seen
>>>quantifiable data either way... if anyone has any,
>>>I'd love to see it.
>>>
>>>RF
>>>
>>>Robert G. Freeman - Oracle8i OCP
>>>Oracle DBA Technical Lead
>>>CSX Midtier Database Administration
>>>
>>>The Cigarette Smoking Man: Anyone who can appease
>>>
>>a
>>
>>>man's conscience can
>>>take his freedom away from him.
>>>
>>>
>>>
>>>-----Original Message-----
>>>Sent: Monday, March 18, 2002 5:54 PM
>>>To: Multiple recipients of list ORACLE-L
>>>
>>>
>>>Robert - So THAT is the title of your next book.
>>>
>>I'm
>>
>>>primed to buy it
>>>already.
>>>I just recalled a legend, maybe. "Disk is 10,000
>>>times slower than memory,
>>>so memory access times are infinitesimal compared
>>>
>>to
>>
>>>disk access". Cary
>>>Millsap covers this in his Hotsos Clinic. He has
>>>
>>run
>>
>>>tests that prove "ain't
>>>so". The point is that you can't just use ratios
>>>
>>to
>>
>>>tune Oracle, but need to
>>>look at wait times.
>>>Dennis Williams
>>>DBA
>>>Lifetouch, Inc.
>>>dwilliams_at_lifetouch.com
>>>
>>>
>>>-----Original Message-----
>>>Sent: Monday, March 18, 2002 1:04 PM
>>>To: Multiple recipients of list ORACLE-L
>>>
>>>
>>>So, does the CoO (Church of Oracle) have an
>>>infallibility doctrine then???
>>>
>>>... From the Book of Oracle, chapter 5 ...
>>>
>>>...and the DBA did look upon his database, and he
>>>saw it was good.
>>>His tablespace datafiles being distributed tither
>>>and fro, spread amongst
>>>the
>>>platters of his disks. And he did complete that
>>>which was called
>>>documentation,
>>>and then he rested from his labors, and drank
>>>Mountain Dew Code Red...
>>>
>>>
>>>:-)
>>>
>>>Robert G. Freeman - Oracle8i OCP
>>>Oracle DBA Technical Lead
>>>CSX Midtier Database Administration
>>>
>>>The Cigarette Smoking Man: Anyone who can appease
>>>
>>a
>>
>>>man's conscience can
>>>take his freedom away from him.
>>>
>>>
>>>
>>>-----Original Message-----
>>>Sent: Monday, March 18, 2002 12:25 PM
>>>To: Multiple recipients of list ORACLE-L
>>>
>>>
>>>
>>>Hey, you're an author!
>>>
>>>I expect perfection, grace and infallibility. ;)
>>>
>>>Jared
>>>
>>>On Monday 18 March 2002 07:33, Freeman, Robert
>>>wrote:
>>>
>>>>And hey, it was Sunday morning at 0700
>>>>
>>>something... what do you
>>>
>>>>expect from me anyways??? :-)))))))
>>>>
>>>>RF
>>>>
>>>>Robert G. Freeman - Oracle8i OCP
>>>>Oracle DBA Technical Lead
>>>>CSX Midtier Database Administration
>>>>
>>>>The Cigarette Smoking Man: Anyone who can
>>>>
>>appease
>>
>>>a man's conscience can
>>>
>>>>take his freedom away from him.
>>>>
>>>>
>>>>
>>>>-----Original Message-----
>>>>Sent: Sunday, March 17, 2002 6:43 PM
>>>>To: Multiple recipients of list ORACLE-L
>>>>
>>>>On Sunday 17 March 2002 07:53, Freeman, Robert
>>>>
>>>wrote:
>>>
>>>>>1. You do not open the database to users until
>>>>>
>>>AFTER you do a backup
>>>(hot
>>>
>>>>>or cold, dosen't mater) at point t2.
>>>>>
>>>>Well, yeah, that was the point. It doesn't have
>>>>
>>>to be a cold backup, but
>>>
>>>>since you can't do any work, it may as well be a
>>>>
>>>cold backup.
>>>
>>>>Jared
>>>>
>>>>>2. There is a method of recovering a database
>>>>>
>>>(8i +) after RESETLOGS has
>>>
>>>>>been
>>>>>issued with archived redo logs. I discussed it
>>>>>
>>>in my DBA World Tour
>>>
>>>>>backup and
>>>>>recovery presentation. To do this, you MUST
>>>>>
>>have
>>
>>>the control file for
>>>the
>>>
>>>>>database from BEFORE the resetlogs operation,
>>>>>
>=== message truncated ===
>
>=====
>Connor McDonald
>http://www.oracledba.co.uk (mirrored at
>http://www.oradba.freeserve.co.uk)
>
>"Some days you're the pigeon, some days you're the statue"
>
>__________________________________________________
>Do You Yahoo!?
>Everything you'll ever need on one web page
>from News and Sport to Email and Music Charts
>http://uk.my.yahoo.com
>
-- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Mogens =?ISO-8859-1?Q?N=F8rgaard?= INET: mln_at_miracleas.dk Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051 San Diego, California -- Public Internet access / Mailing Lists -------------------------------------------------------------------- To REMOVE yourself from this mailing list, send an E-Mail message to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).Received on Wed Mar 20 2002 - 17:03:50 CST