Just to debunk a myth...
Not a technical support thread but an advocacy/opinion debate. Gumby disputes the claim that TPC-C results are driven solely by hardware vendors, citing a Sybase iAnywhere submission, and asks why IBM never publishes TPC-C/TPC-E numbers for Informix. Replies note the Sybase run used only ~$17K of hardware and that meaningful benchmarks cost vastly more, that Oracle/MS (not IBM) ban user-published benchmarks, and suggest the community run and publish its own. Side discussion covers Google's MySQL patches and MySQL's rising share. No resolution; the thread ends in agreement that IBM isn't actively marketing IDS, plus a 'do it yourself' challenge.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Someone (Serge I think...) said that TPC-C benchmarks are the work of the hardware partners. Just look at the icon's and you'll see its *hardware* vendors. Hmmm. Someone had better tell that to Sybase. http://www.tpc.org/tpcc/results/tpcc_last_ten_results.asp That's the link to the TPC Org page. Now if you look, there's a Sybase iAnywhere icon. The benchmark was done 7/29/08. If you click on the icon, you're taken to Sybase, not Dell, the hardware partner. Hmmm. So much for that myth. Now why is it that IM can't get a benchmark of IDS done? I mean heck. If they can improve the product, why can't they get a lousy TPC-C or the promised TPC-E benchmark done? Ya got me. Cause on that note? I don't know ... -G
Ian Michael Gumby wrote: > Someone (Serge I think...) said that TPC-C benchmarks are the work of > the hardware partners. Just look at the icon's and you'll see its > *hardware* vendors. > > Hmmm. > Someone had better tell that to Sybase. > http://www.tpc.org/tpcc/results/tpcc_last_ten_results.asp > That's the link to the TPC Org page. > Now if you look, there's a Sybase iAnywhere icon. The benchmark was > done 7/29/08. > > If you click on the icon, you're taken to Sybase, not Dell, the > hardware partner. > > Hmmm. So much for that myth. > Now why is it that IM can't get a benchmark of IDS done? > > I mean heck. If they can improve the product, why can't they get a > lousy TPC-C or the promised TPC-E benchmark done? > > Ya got me. > Cause on that note? I don't know ... > > -G Gumbi, See also: http://code.google.com/p/google-mysql-tools/wiki/Mysql5Patches | | | Which might lead you to: | | V http://www.bluehost.com/ | | +--> and lead you to--+ | | + the CEO's blog--> http://mattheaton.com/?p=130 | | + http://www.mysqlperformanceblog.com/ I remain curious why, with all this Google interest in MySQL and some other people pushing MySQL into better performance why IBM has said nothing, not even about DB2 performance, vs the stuff coming out for MySQL. Very curious indeed. Surprising not to see any stats or any kind of metrics for DB2, not to mention Informix. Maybe IBM isn't really interested in the database market, instead spending all of its time on moving to India and using MySQL. -- Dutch: What happened to you Dillon, you used to be somebody I could trust! Dillon: I woke up. Why don't you... --From the movie "Predator"
Ian Michael Gumby wrote: > Someone (Serge I think...) said that TPC-C benchmarks are the work of > the hardware partners. Just look at the icon's and you'll see its > *hardware* vendors. > > Hmmm. > Someone had better tell that to Sybase. > http://www.tpc.org/tpcc/results/tpcc_last_ten_results.asp > That's the link to the TPC Org page. > Now if you look, there's a Sybase iAnywhere icon. The benchmark was > done 7/29/08. > > If you click on the icon, you're taken to Sybase, not Dell, the > hardware partner. > > Hmmm. So much for that myth. Umm - the Sybase benchmark hardware cost $17K. I am sure the people on this group could raise $17K if they wanted to. And there's nothing to stop you doing your own benchmark with Informix and getting it audited and published. However, a 20k tpmC result probably ain't going to set the world on fire. "Real" benchmarks (i.e ones that produce numbers that matter) will cost exactly 975.55261761136 times more. Very few software companies are going to spend that much money on a bunch of hardware for a one off.
Mark Townsend wrote: > Ian Michael Gumby wrote: >> Someone (Serge I think...) said that TPC-C benchmarks are the work of >> the hardware partners. Just look at the icon's and you'll see its >> *hardware* vendors. >> >> Hmmm. >> Someone had better tell that to Sybase. >> http://www.tpc.org/tpcc/results/tpcc_last_ten_results.asp >> That's the link to the TPC Org page. >> Now if you look, there's a Sybase iAnywhere icon. The benchmark was >> done 7/29/08. >> >> If you click on the icon, you're taken to Sybase, not Dell, the >> hardware partner. >> >> Hmmm. So much for that myth. > > > Umm - the Sybase benchmark hardware cost $17K. I am sure the people on > this group could raise $17K if they wanted to. And there's nothing to > stop you doing your own benchmark with Informix and getting it audited > and published. However, a 20k tpmC result probably ain't going to set > the world on fire. "Real" benchmarks (i.e ones that produce numbers that > matter) will cost exactly 975.55261761136 times more. Very few software > companies are going to spend that much money on a bunch of hardware for > a one off. Maybe this will help them: http://www.spectra.com/systems/ibm/z9-s54/z9-s54.htm Scroll to the bottom and put us your credit card. <g> -- Daniel A. Morgan University of Washington damorgan@x.washington.edu (replace x with u to respond)
On Sep 16, 12:54 am, Mark Townsend <markbtowns...@sbcglobal.net> wrote: > Ian Michael Gumby wrote: > > Someone (Serge I think...) said that TPC-C benchmarks are the work of > > the hardware partners. Just look at the icon's and you'll see its > > *hardware* vendors. > > > Hmmm. > > Someone had better tell that to Sybase. > >http://www.tpc.org/tpcc/results/tpcc_last_ten_results.asp > > That's the link to the TPC Org page. > > Now if you look, there's a Sybase iAnywhere icon. The benchmark was > > done 7/29/08. > > > If you click on the icon, you're taken to Sybase, not Dell, the > > hardware partner. > > > Hmmm. So much for that myth. > > Umm - the Sybase benchmark hardware cost $17K. I am sure the people on > this group could raise $17K if they wanted to. And there's nothing to > stop you doing your own benchmark with Informix and getting it audited > and published. However, a 20k tpmC result probably ain't going to set > the world on fire. "Real" benchmarks (i.e ones that produce numbers that > matter) will cost exactly 975.55261761136 times more. Very few software > companies are going to spend that much money on a bunch of hardware for > a one off. Now Mark... It goes a bit beyond that. There are relationships involved and of course there's time and having to have your benchmark audited. You know that. Plus if you read the EULA one is not supposed to publish benchmarks that *one* does. (I believe Oracle was one of the first to put these terms in to practice.) Not to mention that its not *my* job to promote nor sell *any* database products. Perhaps thats part of the point I've been making. IBM is relying on the IDS community already to sell IDS, an IBM product on their behalf. Since you're proving to be just as slow on the uptake as most of the IBMers, I'll spell it out to you one more time. IBM is not *selling* IDS and its senior execs are paying lip service to the IDS community. It appears that senior IBM SWG IM management believes that *RDBMS* is a comodity market and not worth the investment. And real benchmarks costs are variable. $17K for hardware, low software cost means a lower cost per tpm. This is one way to judge a benchmark. Another way is to see what 17K wll get you in terms of a constant tpm number. So I can better gauge how much hardware I'll need now and for the next couple of years. As a start up, does it make sense to purchase $250k in hardware/ software or spend much less and put the money in to employess? What's enough hardware /software that keeps me going and will let me focus on my core business? *THATS* the initial point of the benchmarks. But you already knew this. -G
On Sep 16, 12:32 am, InDeep <ind...@indeep.com> wrote: > I remain curious why, with all this Google interest in MySQL and some > other people pushing MySQL into better performance why IBM has said > nothing, not even about DB2 performance, vs the stuff coming out for > MySQL. Very curious indeed. Surprising not to see any stats or any > kind of metrics for DB2, not to mention Informix. Maybe IBM isn't > really interested in the database market, instead spending all of its > time on moving to India and using MySQL. > > - Id, (is there something Freudian with that nickname? ;-) Sorry I digress.... There are some possible reasons for the interest in mySQL. First its open, second, its pretty basic and they can play with it. I'm not sure that Google has any real Database geniuses on staff and those that are near the top seem to want to duke it out with the like's of Stonebraker. The interesting thing is that Google has to create a proprietary database to handle their indexing needs and what they do won't always work best in terms of general purpose databases. I'm interested in why they didn't go postgress instead. But Google is the new Microsoft. Use with caution. -G
Ian Michael Gumby wrote: > On Sep 16, 12:32 am, InDeep <ind...@indeep.com> wrote: > >> I remain curious why, with all this Google interest in MySQL and some >> other people pushing MySQL into better performance why IBM has said >> nothing, not even about DB2 performance, vs the stuff coming out for >> MySQL. Very curious indeed. Surprising not to see any stats or any >> kind of metrics for DB2, not to mention Informix. Maybe IBM isn't >> really interested in the database market, instead spending all of its >> time on moving to India and using MySQL. >> >> - > Id, (is there something Freudian with that nickname? ;-) > > Sorry I digress.... > Not a good idea on this one. > There are some possible reasons for the interest in mySQL. First its > open, second, its pretty basic and they can play with it. They can CHANGE it. Pay attention. > I'm not sure that Google has any real Database geniuses on staff and > those that are near the top seem to want to duke it out with the > like's of Stonebraker. > Google is as Google does. This is something you can snub but it's not going away. Google likes taking free stuff and making other things with it. Their presence alone getting into MySQL should give everyone pause. > The interesting thing is that Google has to create a proprietary > database to handle their indexing needs and what they do won't always > work best in terms of general purpose databases. > > I'm interested in why they didn't go postgress instead. > Do you know why Postgres never got anywhere big like MySQL? Why didn't Ingres ever go anywhere? These two products were on a similar track like Informix, almost with similar features. > But Google is the new Microsoft. Use with caution. > > -G Gum-beeeee. Don't miss the point. MySQL has several reasons to be feared, not to mention Google. One, MySQL _is_ gaining market and mindshare, Informix is not. Two, you brought up KickFire, which was interesting enough on its own terms, and now a benchmark for this basically unknown quantity? This is huge. MySQL zealots can see the light at the end of the tunnel. MySQL brought ease of use first, now it brings performance. These are the same attributes Informix used to have. Whatever happened to SE? It was replaced by MySQL. What will happen to IDS? WE DONT KNOW. If SE is any indication of where Informix will be going, you should be concerned. Snub MySQL all you want, soon enough even Oracle will be paying closer attention. It may well be one of the smartest moves Sun has made in a long time. They bought a ticket to ride. Informix will take yet another back seat to products that are re-inventing the market. -ID-
> Plus if you read the EULA one is not supposed to publish benchmarks > that *one* does. > (I believe Oracle was one of the first to put these terms in to > practice.) Oracle and MS have that restriction in place. I don't believe IBM has for either DB2 or Informix. > Not to mention that its not *my* job to promote nor sell *any* > database products. Then stop complaining > IBM is not *selling* IDS and its senior execs are paying lip service > to the IDS community. We are in agreement (breaks little finger on right hand) > It appears that senior IBM SWG IM management believes that *RDBMS* is > a comodity market and not worth the investment. We are in agreement (breaks little finger on left hand) > And real benchmarks costs are variable. $17K for hardware, low > software cost means a lower cost per tpm. This is one way to judge a > benchmark. Price/performance is a great measure. MS was successful doing this for years. However it does talk to a commodity market, not sure that's where you want to see IDS taken to. > But you already knew this. I'm sort of serious about this. If you guys want to see Informix promoted, then promote it. Publish your own success stories, run your own benchmarks, do your own technical comparisons of IDS vs Oracle, MS and even DB2. Just stop the incessant whining.