Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Linux Business

New Linux TPC-H Record Set 130

prostoalex writes: "New TPC-H world record for performance and scalability of database software on Linux platform has been set. The winner - Oracle 10g running on a four-node Lenovo Cluster Server DeepComp 6800, each with four Intel Itanium 2 1.3 GHz processors. Oracle also emphasizes that it's 3.5 times more performance than similar IBM DB2 benchmark. TPC-H benchmarks are available at TPC Web site."
This discussion has been archived. No new comments can be posted.

New Linux TPC-H Record Set

Comments Filter:
  • Sun is 9th? (Score:3, Interesting)

    by civilengineer ( 669209 ) on Wednesday November 19, 2003 @09:03PM (#7516728) Homepage Journal
    Sun is behind windows by such a huge margin? I thought solaris sets standards for stability.
    • Sun is behind windows by such a huge margin? I thought solaris sets standards for stability.

      Speed is not the same thing as Stability. TPC-H is a test of speed.

      • From a dilbert strip, "Is it ok to do things really fast even if they are really wrong?"

      • Speed is not the same thing as Stability. TPC-H is a test of speed.

        So, not being a db person, I'm curious if there are stability benchmarks?

        Kind of like high TPC * high uptime?

        Just like network vendors can quote error rates like 1 in 10e15 etc., I would imagine that db transactions could be stable like 1 in 10e13, etc. (except for the insurance claim forms I submit that get processed with error rates like 1 in 10e0).

    • Sun boxes havent been fast in a long time hence the move to the AMD chips. Sun just dosent sell enough chips to do the billions in R&D to make a competitive chip. Now beyond the chip sun is great at putting as much IO bandwith as they can arguably second to only SGI without getting into realy esoteric hardware. Even with this said PC's have been getting better and better at IO bandwith with those FSB speeds getting cranked up and AMD with there new new multidirectional FSB.
    • All those huge windows results are made with large shared nothing clusters with basically no redundancy. These are the antithesis of how people run realworld large database systems.
  • Great... (Score:5, Funny)

    by RiffRafff ( 234408 ) on Wednesday November 19, 2003 @09:05PM (#7516740) Homepage
    Now SCO's gonna want $2800 for a license.

    • Re:Great... (Score:4, Funny)

      by Soko ( 17987 ) on Wednesday November 19, 2003 @09:17PM (#7516802) Homepage
      OMG...Imagine if SCO is stupid enough ^W^W^W has the cojones to take on Oracle too. O_O

      "Haha, you fool! You fell victim to one of the classic blunders. The most famous, is never get involved in a litigation war with a company who has more lawyers than Linden, Utah has farmers. But only slightly less well-known is this: Never go in against a meglomaniac when big bucks are on the line! Ahahahahaha! Ahahahaha! Ahahaha--" ~ Larry Elliston

      Soko
      • Re:Great... (Score:1, Offtopic)

        by Dante ( 3418 ) *
        damnit thats funny mod him up!
      • The most famous, is never get involved in a litigation war with a company who has more lawyers than Linden, Utah has farmers.

        While I imagine that involving Ellison would make the PR war a lot more interesting, when it comes to Lawyers SCO has already pretty much chosen to pick on the biggest kid on the block. IBM has a long history of using their (very large) legal department to great effect..

        Still, it might be amusing to watch them try to do this on _even more_ fronts.

  • by mcrbids ( 148650 ) on Wednesday November 19, 2003 @09:09PM (#7516755) Journal
    Linux is clearly being taken seriously. It's pounding the competition in the server space, and it's beginning to make serious inroads to the desktop.

    Desktop Linux stories carry some interest to me, but on a server? That's old hat, old news, and very much humdrum.

    This article really should be more about the cluster of Itanium chips, which actually determine the speed of the system, rather than "it runs Linux!" which in this case is largely irrelevant.

    Linux is as responsible for the success of this as a dog is responsible for the bus that hit it. Similar results could easily be obtained, I'm sure, with any number of BSD variants, or other *nixes compiled to run on Itanium.

    This would have been news 3 years ago, but today? Bah!
    • Linux is as responsible for the success of this as a dog is responsible for the bus that hit it.

      Report to Analogy Class - immediately. HTH. HAND.

      Soko
    • If you look at the page linux indeed scores top in the 100gb and 300gb section. 1000gb it is just behind w2k but in the 3000gb and 10.000gb being talked about in this story. Nada. No linux or windows for that matter.

      So looking at the chart you can only conclude for linux that it rules supreme the 100gb section with double the score of the 2nd. Middle class is really a money choice and since they don't use one currency, how much is a yaun anyway, it is hard to compare. But if you wanna go top neither window

    • by A nonymous Coward ( 7548 ) * on Wednesday November 19, 2003 @10:04PM (#7517018)
      The OS is largely irrelevant to speed tests which never swap or do I/O, like generating graphics. But servers show weaknesses in an OS like nothing else, since they really hammer context switches and I/O.

      This IS significant. It shows the suits that Linux can handle swap intensive tasks, even tho they don't know that is what it shows.
    • I don't think you could replicate this on OpenBSD. Ever tried running MySQL + PHP + Apache on it? Don't. Leave it as your firewall/NAT and let Linux or FreeBSD do the heavy lifting.

    • Desktop Linux stories carry some interest to me, but on a server? That's old hat, old news, and very much humdrum.

      Ummm, it may be old hat and humdrum to you but THE WORD STILL NEEDS TO BE PUT OUT. Just because you're in the minority who is.. tired? of seeing Linux news does not mean the world is tired or should not conintue to see it. The best way to combat non-interest and non-knowledge by the general public is to keep putting the word in front of their noses. Keep saying "See! This is what Linux can do

  • But it is still number 5 on the list. IBM DB2 on a Windows cluster is number 4 and if you are going to go with Oracle then it looks like you really want to be running it on Solaris if performance is your main objective.
    • Or stability for that matter. Oracle 9i RAC clustered on Windows has gotta be the most unreliable db platform you could put together. One of our clients was sending out daily emails reporting server uptime in HOURS, on a database server! That product has been out for 2 years now and it's not stable. I gotta wonder what they had to do to that 10g cluster to make it run. Speed is great, but how about uptime, etc. Win one for Linux, which is good. I hope they improved scalability with 10g as well, as 9i RAC go
  • by kpharmer ( 452893 ) * on Wednesday November 19, 2003 @09:11PM (#7516768)
    are probably comparing this system to some old ibm benchmark. They didn't say in the press release, so I'd assume the worst.

    IBM appears to dominate the TPC-Hs at the top & bottom, with oracle owning it in the middle.

    The only really interesting benchmark out there at the moment is the IBM DB2 ICE configuration - in which they spread db2 across dozens of low-end AMD Opteron dual-cpu servers. DB2 (and informix god bless them) partition differently than oracle - more like a database implementation of beowulf (that they've been doing for 8+ years). Way cheaper than anything from oracle, and you can toss up to 1000 servers into it. Their benchmark is in the 300 gbyte range, not 1000 - but it'll scale way beyond oracle, and is cheap for that kind of power: http://www.tpc.org/tpch/results/tpch_result_detail .asp?id=103073001

    Makes me wonder how many pcs I've got laying around the house...
  • by Preach the Good Word ( 723957 ) on Wednesday November 19, 2003 @09:11PM (#7516771)
    there are basically three type of clusters:

    1) shared nothing: in this, each computer is only connected to each other via simple IP network. no disks are shared. each machine serves part of data. these cluster doesn't work reliably when you have to aggregations. e.g. if one of the machine fails and you try to to "avg()" and if the data is spread across machines, the query would fail, since one of the machine is not available. most enterprise apps cannot work in this config without degradation. e.g. IBM study showed that 2 node cluster is slower and less reliable than 1 node system when running SAP.

    IBM on windows and unix and MS uses this type of clustering (also called federated database approach or shared nothing approach).

    2) shared disk between two computers: in this case, there are multiple machines and multiple disks. each disk is atleast connected to two computers. if one of the computer fails, other takes over. no mainstream database uses this mode, but it is used by hp-nonstop. still, each machine serves up part of the data and hence standard enterprise apps like SAP etc cannot take clustering advantage without lot of modification.

    3) shared everything: in this, each disk is connected to all the machines in the cluster. any number of machines can fail and yet the system would keep running as long as atleast one machine is up. this is used by Oracle. all the machine sees all the data. standard apps like SAP etc can be run in this kind of configs with minor modification or no modification at all. this method is also used by IBM in their mainframe database (which outsells their windows and unix database by huge margine). most enterprise apps are deployed in this type of cluster configuration.

    the approach one is simpler from hardware point of view. also, for database kernel writers, this is the easiest to implement. however, the user would need to break up data judiciously and spread acros s machines. also adding a node and removing a node will require re-partitioning of data. mostly only custom apps which are fully aware of your partitioning etc will be able to take advantage.
    it is also easy to make it scale for simple custom app and so most of TPC-C benchmarks are published in this configuration.

    approach 3 requires special shared disk system. the database implementation is very complex. the kernel writers have to worry about two computers simultaneously accessing disks or overwriting each others data etc. this is the thing that Oracle is pushing across all platforms and IBM is pushing for its mainframes.

    approach 2 is similar to approach 1 except that it adds redundancy and hence is more reliable.
    • Actually, Microsoft employs both federated and failover clustering, and the two are not mutually exclusive - you can build a federated cluster with failover nodes - to one another or to a hot spare. Federating and failover aren't really related. Federation is a way of dividing up your large tables across multiple database servers for performance, failover is for redundancy.

      most enterprise apps are deployed in [the shared everything] type of cluster configuration.

      Really? Wow. You sure about that? I would
  • oracle and linux (Score:3, Insightful)

    by larry bagina ( 561269 ) on Wednesday November 19, 2003 @09:12PM (#7516775) Journal
    Oracle walks a dangersous path running on linux. Sure, the money saved by using linux/x86/oracle vs solaris/SPARC/oracle is significant, but linux can be a gateway drug to other Open Source/FREE software. Once PHBs realize that the OS is a commodity, the next step is realizing the DB is also a commodity. Postgresql or mysql isn't suitable for enterprise-level work, but it's more than suitable for small internal projects that used to mean extra orcle seats.
    • by Anonymous Coward

      Postgresql or mysql isn't suitable for enterprise-level work

      MySQL certainly isn't, but PostgreSQL can do anything Oracle can do. The only things it lacks are mindshare and access-style forms. This doesn't stop it from running the .org DNS registry, or many other "enterprise-level" tasks, and several projects are working on the access-style forms, including Kexi, PGAdmin3, and the newly liberated Rekall. Mindshare is a little more difficult, but given time, it will come.

      • > but PostgreSQL can do anything Oracle can do

        oh please, you don't really believe that do you?

        I mean, come on - postgesql is a great little database with a bright future...but it isn't ready to tackle large ERP or CRM implementations yet. Or even your typical 300 gbyte warehouse or data mart if you get right down to it...
        • Re:oracle and linux (Score:3, Informative)

          by alannon ( 54117 )
          Apparently the American Chemical Society has a Postgresql database in use that's over a terabyte in size. I don't know if this is the largest one currently in use.

          Also the largest commercial database is about 23Tb and runs on Oracle.

          What these numbers don't say anything about, though, is how much of these databases are taken up by BLOBs, and how much is actual field data. Having most of your data in BLOBs is really just making your database a fancy file system, since BLOBs reside in a different part of
    • by nemesisj ( 305482 )
      This is not an insightful post. Serious people use Oracle. Poor people use PostgreSQL. If you need Oracle, you will have to use it or DB2 - PostgreSQL just can't cut it on several levels with Oracle, currently.
      • Serious people use Oracle.

        Well not me - Oracle shafted me when Oracle for DOS was discontinued and my enterprise's direct mailing system was based on it. I was shafted by Oracle for OS/2 being discontinued when the inventry management system used in my enterprise was based on it, and shafted when Oracle Power Objects used for my factory management system died from Y2K problems.

        I chose PostgreSQL over Oracle for my enterprise - Open source cannot die on you, be withdrawn, or have support withdrawn.

        I wou

        • Serious people use Oracle.

          I'm sorry, but did serious people ever use Oracle for DOS? [Or Oracle for OS/2 for that matter?] I have done consultancy at many (>100) Oracle sites over the years and am yet to see either of those configurations being used in anger. Maybe you need to look at your organisation's OS selection policy before being so quick to blame companies like Oracle for pulling the plug...

          I'm no great fan of Oracle. There IS lots wrong with Oracle. But be fair... pulling support for Oracle

    • As somebody who worked for O$ for many years, I'm interested by Larry's jump into bed with Linux. I'm out of the loop now, but I wouldn't be surprised if this is part of a "testing the OSS water" strategy. I also wouldn't be surprised to see some low revenue Oracle products (e.g. JDeveloper, Developer, Application Server) being turned open source.

      "...but linux can be a gateway drug to other Open Source/FREE software"

      For now, I don't think this is a (serious) risk. Oracle has been distributing Apache now

  • by Qrlx ( 258924 ) on Wednesday November 19, 2003 @09:13PM (#7516776) Homepage Journal
    Shouldn't that read "New TPC-H Record Set Using Oracle?"

    The article didn't give much details, but how much of this performance is directly attributable to Linux (specifically Red Hat AS3)? What was the OS of the system it beat? Could that also have been Linux? How much of the performance can be attributed to the (suspiciously un-Beowulf) Lenovo cluster?

    From what I know of benchmarks, the numbers given reflect real-world preformance, to within one order of magnitude.

    At first, I thought, It's just a press release, big deal... But wait, they used Linux, so it must be another straw on the back of the camel knows as the Closed Source Business Model. But wait, it's running Oracle, so it must therefore be evil. Aieeeeeeeeeeeeee!!!!
    • I don't know how much of it is attributable to linux (probably not much). The bigger story is that linux is able to scale and run real world enterprise databases without loss of performance or stability.

      This is important to counter MS FUD. Today Bill Gates said Linux is what Unix was in the 1970s: a perfectly reasonable operating system. [com.com]. Articles like this make it possible for other people (say Red Hat for example) to say that Bill Gates is full of shit.
    • Shouldn't that read "new TPC-H Record Set Using Itanium2?" These "benchmarks" are hogwash, they don't measure one vendor's database against the others in general. They measure overall solution where hardware plays the key role. And they make good proof of concepts. Not to mention other politics (and discounts) in the mix... Vendors need to pick and choose their benchmarks so as to not outdo their other strategic benchmarks. The only interesting benchmark on that whole website in respect to the db vendor a
  • by Fnkmaster ( 89084 ) * on Wednesday November 19, 2003 @09:13PM (#7516778)
    "Myaaa, Did you get the memo? We're now using the new cover on all TPC Reports. If you could just do that, that would be great. Thanks."
  • Looks like linux is moving up in the world.
    I'm still impressed by teradata.
    But what is MP-RAS 3.02.00 OS that the 3,000 GB Results on Terra Data ran on?
  • Linux is almost double that of the MS solutions or solaris solutions when you compare price per QphH. Anyway this for datawarehousing, real test is the C which is relational database tests.
  • Anyone else read that as THC??
  • by Hoser McMoose ( 202552 ) on Wednesday November 19, 2003 @09:35PM (#7516887)
    Holy crap this story is useless! Go to the TCP-H site and actually look at the results, it really is nothing even remotely impressive.

    - It's NOT the fastest TCP-H result, it's the fastest LINUX TCP-H 1000GB result. Actually it's the ONLY Linux TCP-H 1000GB result. 5th of 8 overall

    - It's not even offering very good bang for your buck, coming in 5th of 8 for Price/QphH ($156 US according to today's currency exchange). The only systems it managed to beat are two outdated systems (both from HP) and an old price for a Fujitsu system, quoted in euro (the same system offers the same performance but a lower price on a newer entry quoted in US $).

    In short, if anything this suggests that Linux is a BAD choice for this work! The performance isn't there and the cost is high.

    Where things get REALLY bad though is the claim that this is "3.5 times faster" than a system running IBM's DB2. This is just 100% pure bullshit! The new Linux/Oracle system runs 1.3GHz Itanium2 processors and Oracle 10g. The HP/Windows/DB2 system runs 900MHz Xeon processors and runs DB2 7.2 (8.1 is current version). What's more, the Oracle/Linux system isn't even 3.5 times faster, it's just 3.5 times faster PER PROCESSOR! Great, your brand-spanking new Itanium2 is 3.5 times faster than four year old Xeon 900MHz chips. Whoopie!

    Note: if you do want to see impressive Linux results, look at what IBM is doing with their Opteron cluster and DB2 running under SuSE Linux. They turned in the top results in the two TPC-H tests they entered (100GB and 300GB).
    • It's not even good price per unit (Sun wins on the cost per unit in the lower sized tests, 100 GB & 300 GB respectively are $28 & $28 US), this result was nearly $160 US per unit once you do the currency conversion. That puts this result at the 4th most expensive per unit it its class, and smack in between the two most powerful raw solutions in its cost per unit, both of them being Fujitsu Sparc running Solaris, one being $185 US per unit and the other being $141.

      Someone remind me why Sun, Sparc, a

    • by Anonymous Coward
      have to point out the Xeon processors have 2Mb L2 cache, where as the the Itanium2 have 512kb. That makes a huge difference for TPC-H queries. Plus, don't believe Intel's hype about Itanium. Xeon is still a kick ass CPU.
  • by Anonymous Coward
    SGI have built the largest Linux machine (512 processor machine at NASA [sgi.com]) and managed to destroy the previous memory bandwidth record [sgi.com] held by NEC, by achieving 1 terabyte/s.
    • Wow, talk about using a NARROW definition of "largest Linux machine". There are larger Linux systems, some using thousands of processors in a cluster. There are also other companies out there building systems that can have up to 64 processors per node like this SGI system. But SGI is using a combination of software and hardware to get global shared memory instead of a more traditional cluster design.

      Of course, Linux's NUMA optimizations are still in their infancy (they are pretty much non-existant in th
      • by Anonymous Coward
        No, its the largest Linux machine. Not a cluster, but a coherent, single system image.

        Obviously SGI had to modify the 2.4 kernel to achieve good performance and use extensive NUMA modifications. No, "using NUMA over internode communication link" doesn't kill your scalability. NASA isn't that stupid, neither is SGI.

        The guys at SGI are currently testing the linux 2.6 kernel with a 512 processor system. They are saying (unsurprisingly) that it looks like it is a great improvement over 2.4.
  • by jjeffries ( 17675 ) on Wednesday November 19, 2003 @09:59PM (#7516987)
    I was wondering who bought all the Itanium 2s....
  • So many zealots from both PostgreSQL and MySQL sides are publishing their thoughts on which OSS DBMS is faster, but I do not see any high-end test results from them? Why TPC results do not include anything from OSS DBMSs? No results or nobody cares or OSS DBMS cannot have any TPC results by some political reasons? Can someone explain?
    • I thought it was widely known that MySQL is made for speed, and postgres is made for reliability.

      • You thought wrong: on complex queries PostgreSQL is fatser AND still more reliable.

        But the point is not what you or me thought. The point is to see such benchmark results on the official TPC site. Or, if it (the space on that site) is expensive then somewherelese - but it must be based on TPC,

    • MySQL? (Score:1, Flamebait)

      by autopr0n ( 534291 )
      I'm not even sure if MySQL is even capable of running these tests, what with needing transactions and all. Do they have that yet?
    • PostgreSQL and MySQL can't be active-active clustered (yet). They've only recently had replication master-slave(s) added. But stay tuned, I'm sure it's coming.....and then the open source databases will eat Oracle's lunch.
      • So, no TPC-H then yet. But how about TPC-C? That test doesn't require clusters, AFAIK.

        More generally, I'd like to see the cost/performance comparison of PostgreSQL vs MySQL vs Oracle vs DB2 vs MS SQL 2000 for several types of applications, similar or the same as in TPC-C or TPC-D.

        • sure I could do TPC-C for one to n threads for Oracle and PostgreSQL, but my machines are so very old no one would quote or use them (who wants to see TCP-C on a 500MHz Celeron with 256M of RAM for Oracle 8i?). MySQL might be a problem since the usual table type everyone uses (MyISAM) doesn't even support transactions.......
  • Better summary (Score:4, Insightful)

    by autocracy ( 192714 ) <slashdot2007@sto ... m ['mo.' in gap]> on Wednesday November 19, 2003 @10:07PM (#7517034) Homepage
    Sun Micro kicked everybody's ass. Read across the board, they had the cheapest cost per performance and though Fujitsu systems really shined through on the 1000GB test, they're still SPARC architecture and still running Solaris.

    Truthfully, I'm not a Sun fanboy (I just think they make cool shiny toys that cost a lot). Despite their corporate issues of late, they can still flex when it comes time to move things. Given any of those system built into a decent cluster (note that no pure Sun solutions were clustered), I think something worthwhile might show up.

    Even if you disagree with me on those points though, you do have to agree that the /. article itself just sucked.

  • by RalphBNumbers ( 655475 ) on Wednesday November 19, 2003 @10:10PM (#7517051)
    As you can see here [tpc.org], the DB2 systems they seem to be comparing themselves with scored more than double what this one did.

    I would expect a larger system to score lower on a pre-processor basis just from scaling issues, even if the processors were identical.

    While the 3.5x ratio is impressive, the manner of it's announcement is very misleading.
  • Am I missing something but wasn't it IBM and Redhat and now Novell and Suse? If you look at the top of the 10gb and 100gb sections then this makes for some weird partners. Can't Redhats software be used for some reason with DB2?

    For other odd results, Oracle doesn't seem to go up as high as DB2 and Teradata. I thought Oracle was supposed to be the heavy hitter?

    Anyone got a clue as to how the opensource databases would fare in these tests?

    • IBM has a strong commitment to SLES and Redhat. Actually, there is a stronger bond with SLES among the developers and testers. Suse is much quicker to respond to defect/bugzilla reports than Redhat.

    • To answer your question, I am not sure if you can cluster MySQL but it beats the pants off Oracle if you lay out your data intelligently. However, MySQL doesn't support read consistency, rollbacks, or the concept of a transaction in super speedy mode. I use MySQL for quick data analysis. MySQL is easy to setup, is light on resources, and can load data like a banshee. Oracle is not so speedy to setup and is a resource pig, however I trust Oracle with all my important data since it's a very solid product
  • by Anonymous Coward
    and bitch about how the MS solution is better, here is a little secret. If you look at the current #3 from HP http://www.tpc.org/tpcc/results/tpcc_result_detail .asp?id=103082701 [tpc.org], you see it says COM+. Well that's not the whole truth. If you look at the actual source code, you will see references to tuxedo. It's a C++ port of tuxedo. the original TUX/TUXEDO was created by AT&T http://www.middleware.net/tuxedo/articles/tuxedo_h istory.html. Microsoft isn't stupid, but it's hardly surprising. It doesn't
    • No experience with Unix?

      Hmm they even wrote one years ago. 1980 to be exact. Some of us even remember our xenix floppies.

      How about Xenix which could be argued quite well that it morphed into Sco.

      MS has quite a bit of experience with Unix do not fool yourself. Even ported IE 4.0 over way back in the day.

      Though MS has some evil qualities, to not overlook the fact they have many smart people working there. And they are not stupid.

      Puto
  • No cheer leading from me...

    Seriously, isn't Oracle involved in the Great Firewall of China?

    What are they storing, a list of all the sites that AREN'T allowed to be viewed?
    A list of the 1 billion hotmail addresses used by mainland Chinese?

    It is nice that they've pushed the tech to a new level, but you've got to think...
  • Exchange Rate (Score:3, Insightful)

    by charnov ( 183495 ) on Wednesday November 19, 2003 @11:56PM (#7517545) Homepage Journal
    Sorry, the only reason that this resulted in a new record was because of the artifically controlled exchange rate between the Yuan and the dollar.

    Sorry, try again.
  • ...I really care about database app benchmarks...
  • "The TPC Benchmark(TM)H (TPC-H) is a decision support benchmark." i.e. for management accountants.
    "The TPC-C benchmark continues to be a popular yardstick for comparing OLTP performance on various hardware and software configurations." i.e. for me to get cash from an ATM

    There's only one <a href="http://www.tpc.org/tpcc/results/tpcc_result _ detail.asp?id=103090501">result</a> for TCP-C, which looks OK but not stunning. The <a href="http://www.tpc.org/results/FDR/TPCC/HP%20Int egrity%20rx567
  • So let me get this straight. NASA has installed a big new Linux-based supercomputer, consuming only 5.2kW of power per rack, that has broken previous transaction processing benchmarks running Oracle 10g.

A Fortran compiler is the hobgoblin of little minis.

Working...