Re: Slightly-OT: Throw HW at a SW/DB problem
Date: Wed, 29 Jun 2011 07:45:00 -0700 (PDT)
Message-ID: <1309358700.9350.YahooMailRC_at_web113214.mail.gq1.yahoo.com>
>> Isn't that the whole point of Exadata/logic as it is being promoted nowadays?
>> (the first one to reply with "Exa-whatever is a lot more than just hardware"
>>gets the dumbass award of the year...)
The following represents my opinion and in no way represents the opinion of Oracle Corporation......
Wow, now that's just a pretty tough statement to make. Exadata IS more than just hardware. Try getting HCC to work in your non-exadata database Nuno.... Try offloading your SQL Queries to your (essentially) federated storage for parallel processing at the storage level... Granted, Exadata is some big iron, but it's big iron with a purpose beyond trying to make a slow application faster. Sure, the upshot of the whole hardware and software combination is that it DOES so. The point is that with respect to things like consolidation, reducing costs (ie: electricity, cooling, footprint), the cloud, hardware refreshes, etc.... Exadata is a compelling story. When you are looking at "big data" and you look at competitors in the massive data market, Exadata can do the same work with fewer machines, at a cheaper overall price.
The point is that we are not promoting Exadata as a hardware fix for bad code (though by default it becomes that and we do say it will make your database run faster). We promote it for a number of reasons beyond that. At the end of the day, the preference would still be to actually fix the bad code even on the Exa* platform. But.... when you have a 10 or 20 year investment in your legacy code, and the performance problem has occurred slowly, over time, due to increasing data volumes and increasing concurrency, then there needs to be solutions that are short term and longer term. Your not going to fix legacy code in an instant, and you can't wait around for reports to run over the next 48 hours. If you have a vendor who writes their code in some other database and then "ports" it over and it performs poorly in Oracle and the vendor does not really have Oracle resources to throw at the problem, just what are you going to do (yes yes, profiles and all that, I know).
I've seen more than one case where the legacy code pre-exa* had become so slow over time that it was becoming impossible for the business to actually run. Reports were not completing, etc... Put it on Exa* and now all of a sudden a 48 hour process completes in less than an hour. Now, one might say that there was some lack of foresight (they didn't see and plan for the data volumes) but so what? Do you stop business for two years for a major rewrite, schema re-architect? Do you stop business for 2 years so you can impliment something else? How many solutions are truly "one-shot" solutions that fix everything? What about the simple growth in concurrency that was not foreseen 10 years ago? Do we stop the business for a couple of years to re-architect the whole thing so it can handle 1000 concurrent users rather than the 200 planned for? How much will that cost and how long will it take?
Exadata has a lot of stories out there, and the problem of "fixing the code" seems to have gotten this Voldermort like mysticism and heaven help anyone who objects to the notion that there may be other reasons that more hardware is appropriate......
The day that you can collect all the CERN data from one collier experiment running Oracle on my laptop is the day that I will accept that hardware is not sometimes (even often) a viable solution to a specific problem.
So now, you may award me the dumbass of the year award......
Robert G. Freeman
Master Principal Consultant, Oracle Corporation, Oracle ACE
Author of various books on RMAN, New Features and this shorter signature line.
Blog: http://robertgfreeman.blogspot.com
Note: THIS EMAIL IS NOT AN OFFICIAL ORACLE SUPPORT COMMUNICATION. It is just the opinion of one Oracle employee. I can be wrong, have been wrong in the past and will be wrong in the future. If your problem is a critical production problem, you should always contact Oracle support for assistance. Statements in this email in no way represent Oracle Corporation or any subsidiaries and reflect only the opinion of the author of this email.
From: Nuno Souto <dbvision_at_iinet.net.au> To: oracle-l_at_freelists.org
Sent: Wed, June 29, 2011 1:27:20 AM
Subject: Re: Slightly-OT: Throw HW at a SW/DB problem
Rich Jesse wrote,on my timestamp of 28/06/2011 12:45 AM:
>http://www.sqlmag.com/article/Performancetuning/ssds-performance-tuning-experts-milkman-139591
>1
Oh boy! Another irresponsible twerp parading as a "seasoned professional". Th eonly "seasoning" this idiot demonstrates is a total lack of any industry experiece "salt"...
>
> Anyone from the Oracle camp advocating/predicting hardware-based "tuning"?
Isn't that the whole point of Exadata/logic as it is being promoted nowadays?
(the first one to reply with "Exa-whatever is a lot more than just hardware" gets the dumbass award of the year...)
> Rant over.
Ibidem.
- Cheers Nuno Souto in wet Sydney, Australia dbvision_at_iinet.net.au -- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-l