DB2 DOUBLES THE PERFORMANCE OF ORACLE 10G
Not a support question but a marketing/benchmark flame thread: someone posted an IBM page claiming DB2 doubles the performance of Oracle 10g RAC on a 1TB TPC-H test. Posters picked it apart, noting the comparison used different CPUs (1.7GHz POWER4+ with 128MB L3 vs 1.3GHz Itanium2), different operating systems, and that IBM's graph omitted three higher 1TB results listed on tpc.org. Side discussions covered XPS lacking TPC-friendly features (e.g. materialized views) and shared-nothing vs shared-everything failover. Ends in IBM/Oracle sniping with no conclusion or resolution.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
"Tim Schaefer" <tim@spamethnot.com> wrote in message news:UvdZb.83044$Ct5.51204@fe06.usenetserver.com... > http://www-306.ibm.com/software/data/db2/benchmarks/121503.html > > ( Gawd-I-just-love-this-shit :-) > > > IBM DB2 UDB Enterprise Server Edition v8.1 on 4 IBM eServer p655's ( 4 x IBM Power 4 1700 MHz) running IBM AIX 5L V5.2; Oracle Database 10g Enterprise Edition with Real Application Clusters on 4 Legend DeepComp 6800 Server ( 4 x Intel Itanium2 1.3GHz) running Red Hat Linux Advanced Server v3 for Itanium Yeah right, 4x1700Mhz vs 4x1300 Mhz. Different CPU architectures Different Operating Systems. Like we can compare that! Oracle: Processors: 16 x Itanium2 1.3GHz, 32KB L1-cache, 256KB L2-cache, 3MB L3-cache IBM: Processors 4 x 1700MHz POWER4+ with 128MB L3 Cache So Oracle has 48Mb cache on the CPU and IBM has 128Mb! Try again..this time for the same OS...On the same hardware! Then we can take the pi*s! You're not going to throw a strop at me are you?
http://www-306.ibm.com/software/data/db2/benchmarks/121503.html ( Gawd-I-just-love-this-shit :-)
Tim Schaefer wrote: > http://www-306.ibm.com/software/data/db2/benchmarks/121503.html > > ( Gawd-I-just-love-this-shit :-) I'd love to see an XPS benchmark, wouldn't you? -- "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" - Coluche
>> DB2 Doubles the Performance of Oracle 10g RAC That's an amazing trick, isn't it? So you buy Oracle 10g, then just buy DB2, and magically your Oracle 10g performance doubles?
Obnoxio The Clown wrote: > Tim Schaefer wrote: > > >>http://www-306.ibm.com/software/data/db2/benchmarks/121503.html >> >>( Gawd-I-just-love-this-shit :-) > > > I'd love to see an XPS benchmark, wouldn't you? > I'd love it. But I suppose XPS currently lacks some goodies very insteresting for TPC... Anyway... does anybody, except the deci$sion makers (!?!?!), care about TPC? We the techies all know they're just bull**** don't we?! ;)
----- Original Message ----- From: "Fernando Nunes" <spam@domus.online.pt> Newsgroups: comp.databases.informix Sent: Friday, February 20, 2004 10:22 AM Subject: Re: DB2 DOUBLES THE PERFORMANCE OF ORACLE 10G > Anyway... does anybody, except the deci$sion makers (!?!?!), care about TPC? We the techies all know they're just bull**** don't we?! ;) ... except of course when they reinforce options we already held!
Fernando Nunes wrote: > Obnoxio The Clown wrote: >> Tim Schaefer wrote: >> >> >>>http://www-306.ibm.com/software/data/db2/benchmarks/121503.html >>> >>>( Gawd-I-just-love-this-shit :-) >> >> >> I'd love to see an XPS benchmark, wouldn't you? >> > > I'd love it. But I suppose XPS currently lacks some goodies very > insteresting for TPC... Like what? -- "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" - Coluche
Nah... XPS was cool... but it's not relevant anymore. Too bad. Great set of tools for that one. Keeping in mind who DB2 has to really compete against, those bench marks are sweet! "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message news:c14g54$1e8c0p$1@ID-64669.news.uni-berlin.de... > Tim Schaefer wrote: > > > http://www-306.ibm.com/software/data/db2/benchmarks/121503.html > > > > ( Gawd-I-just-love-this-shit :-) > > I'd love to see an XPS benchmark, wouldn't you? > > -- > "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" > - Coluche
My thoughts exactly... You'll notice I took the words right off the web site. :-) "Captain Pedantic" <theharlequin36@hotmail.com> wrote in message news:c14ig7$1e17qe$1@ID-162943.news.uni-berlin.de... > >> DB2 Doubles the Performance of Oracle 10g RAC > > That's an amazing trick, isn't it? > So you buy Oracle 10g, then just buy DB2, and magically your Oracle 10g > performance doubles? > >
Tim Schaefer wrote: > My thoughts exactly... You'll notice I took the words > right off the web site. :-) > > "Captain Pedantic" <theharlequin36@hotmail.com> wrote in message news:c14ig7$1e17qe$1@ID-162943.news.uni-berlin.de... > >>>>DB2 Doubles the Performance of Oracle 10g RAC >> >>That's an amazing trick, isn't it? >>So you buy Oracle 10g, then just buy DB2, and magically your Oracle 10g >>performance doubles? >> >> No joke, we heard of instances where customers ran Oracle through DB2 Datajoiner do get better plans. Now that's a few years back. Mere anecdotal value.... -- Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab
Serge Rielau wrote: > No joke, we heard of instances where customers ran Oracle through DB2 > Datajoiner do get better plans. Now that's a few years back. Mere > anecdotal value.... > I'll think you find that it was an IBM test, not a customer one, and highly contrived. Basically DJ would pull the data back and hash join, which was faster than the NLJ being done in Oracle. And Oracle was not doing hash joins, because, believe it or not, in the example scenario, the database had been set up not to do them. Wonder why ? The IBM web page is also a little selective in the benchmarks it's using for comparison. Check out the rest of the 1 TB results at http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=all There are actually three higher results NOT show on the graph (and I wonder who has them) > Gawd-I-just-love-this-shit I do believe Tim's categorization is right.
Tim Schaefer wrote: > http://www-306.ibm.com/software/data/db2/benchmarks/121503.html > > ( Gawd-I-just-love-this-shit :-) Ignoring the obvious differences in hardware making this claim a obvious cheap shot from a marketing department that can't put up a 1:1 comparison because it would have nothing to point to ... What happens when you pull the plug on one of those shared nothing nodes as compared to one of Oracle's shared everything nodes? What is the throughput? Oh yeah I remember ... shared nothing architecture has no hardware failures. Hardware failures only happens to the other guy. -- Daniel Morgan http://www.outreach.washington.edu/ext/certificates/oad/oad_crs.asp http://www.outreach.washington.edu/ext/certificates/aoa/aoa_crs.asp damorgan@x.washington.edu (replace 'x' with a 'u' to reply)
"Mark Townsend" <markbtownsend@comcast.net> wrote in message news:p3qZb.23539$4o.41638@attbi_s52... > Serge Rielau wrote: > > The IBM web page is also a little selective in the benchmarks it's using > for comparison. Check out the rest of the 1 TB results at > http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=all > There are actually three higher results NOT show on the graph (and I > wonder who has them) Ha! Ha! Ha! Great spot!
"Daniel Morgan" <damorgan@x.washington.edu> wrote in message news:1077297279.55385@yasure... > Oh yeah I remember ... shared nothing architecture has no hardware > failures. Hardware failures only happens to the other guy. Certainly not people using the top-ranking Siemens hardware. Siemens have a great reputation over here in England-land for the quality of their government IT projects. Who remembers the eight-hour queues for a passport of a couple of summers ago ... ?
Mark Townsend wrote: > Serge Rielau wrote: > >> No joke, we heard of instances where customers ran Oracle through DB2 >> Datajoiner do get better plans. Now that's a few years back. Mere >> anecdotal value.... >> > I'll think you find that it was an IBM test, not a customer one, and > highly contrived. Basically DJ would pull the data back and hash join, > which was faster than the NLJ being done in Oracle. And Oracle was not > doing hash joins, because, believe it or not, in the example scenario, > the database had been set up not to do them. Wonder why ? This is the second time in a couple of weeks where you seem to be certain to know what I talk about. In neither case did you hit it on the nose. The other case was the >100 node DB2 system, alleged to be a contrived, partitioned OLTP system. > > The IBM web page is also a little selective in the benchmarks it's using > for comparison. Check out the rest of the 1 TB results at > http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=all > There are actually three higher results NOT show on the graph (and I > wonder who has them) Well.. one has to be more than a little selective to find a RAC result. Why do I have the feeling that two of them are dups? That each of them uses 4 times the number of CPU's to get 50% more juice, and none of them uses RAC, but instead a big SMP box. I recently saw an ad on that topic....onDemand vs Grid. Can't quite put my finger on it. Cheers Serge Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab
Obnoxio The Clown wrote: >>I'd love it. But I suppose XPS currently lacks some goodies very >>insteresting for TPC... My knowledge on XPS is near null... I was thinking about materialized views, and sometime ago I've read or heard about a DB2 feature where one query could hook up to an already running query (similar or identical queries...?) Ok... this last one seems weird... I'm confused... forget it. Regards.
Daniel Morgan wrote: > Tim Schaefer wrote: > >> http://www-306.ibm.com/software/data/db2/benchmarks/121503.html >> >> ( Gawd-I-just-love-this-shit :-) > > > Ignoring the obvious differences in hardware making this claim > a obvious cheap shot from a marketing department that can't put > up a 1:1 comparison because it would have nothing to point to ... > > What happens when you pull the plug on one of those shared nothing > nodes as compared to one of Oracle's shared everything nodes? What > is the throughput? > > Oh yeah I remember ... shared nothing architecture has no hardware > failures. Hardware failures only happens to the other guy. You throw in another 4way in standby....oh yes.. you never accepted that proposition in the past but never explained why. Why? Cheers Serge -- Serge Rielau DB2 SQL Compiler Development IBM Toronto Lab
Tim Schaefer wrote: > Nah... XPS was cool... but it's not relevant anymore. Too > bad. Great set of tools for that one. Keeping in mind who > DB2 has to really compete against, those bench marks are sweet! Yes, the comparison of apples to pears always is. > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > news:c14g54$1e8c0p$1@ID-64669.news.uni-berlin.de... >> Tim Schaefer wrote: >> >> > http://www-306.ibm.com/software/data/db2/benchmarks/121503.html >> > >> > ( Gawd-I-just-love-this-shit :-) >> >> I'd love to see an XPS benchmark, wouldn't you? -- "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" - Coluche
Mark Townsend wrote: > Serge Rielau wrote: > >> No joke, we heard of instances where customers ran Oracle through DB2 >> Datajoiner do get better plans. Now that's a few years back. Mere >> anecdotal value.... >> > I'll think you find that it was an IBM test, not a customer one, and > highly contrived. Basically DJ would pull the data back and hash join, > which was faster than the NLJ being done in Oracle. And Oracle was not > doing hash joins, because, believe it or not, in the example scenario, > the database had been set up not to do them. Wonder why ? > > The IBM web page is also a little selective in the benchmarks it's using > for comparison. Check out the rest of the 1 TB results at > http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=all > There are actually three higher results NOT show on the graph (and I > wonder who has them) > >> Gawd-I-just-love-this-shit > > I do believe Tim's categorization is right. I must have just taken leave of my senses (well, alright, again.) I could have sworn someone from Oracle was claiming the moral high ground in the benchmark war and sounding aggrieved in the process...? -- "C'est pas parce qu'on n'a rien ''' dire qu'il faut fermer sa gueule" - Coluche
Serge Rielau wrote: > This is the second time in a couple of weeks where you seem to be > certain to know what I talk about. Third, but, hey, who's counting.