Path: dp-news.maxwell.syr.edu!spool.maxwell.syr.edu!drn.maxwell.syr.edu!news.maxwell.syr.edu!border1.nntp.dca.giganews.com!nntp.giganews.com!local01.nntp.dca.giganews.com!nntp.comcast.com!news.comcast.com.POSTED!not-for-mail NNTP-Posting-Date: Fri, 04 Nov 2005 05:59:53 -0600 From: "VC" Newsgroups: comp.databases.theory References: <1130610162.744817.239070@g14g2000cwa.googlegroups.com> <1130765800.098046.93870@f14g2000cwb.googlegroups.com> <9v-dnYE-SanY6PreRVnygg@pipex.net> <1130939293.784399.319040@g43g2000cwa.googlegroups.com> <1130949100.933681.55190@z14g2000cwz.googlegroups.com> <1130965639.482009.182400@g49g2000cwa.googlegroups.com> <2eyaf.8082$Hj2.1929@news-server.bigpond.net.au> Subject: Re: Modelling objects with variable number of properties in an RDBMS Date: Fri, 4 Nov 2005 07:00:03 -0500 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2900.2180 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 X-RFC2646: Format=Flowed; Response Message-ID: Lines: 57 NNTP-Posting-Host: 24.91.127.140 X-Trace: sv3-jGGQzp4+GlV2hdJ+6byHmMmSa/GtHtSz+1B6p6fLmfO0DPUi+CY17Jjc2QdhcBjRhEEC4HzaUWI/eKR!l+M7zK23mLx8NnndezjDSfMuv4RaeqOzVx6h/lm1/co154teR2nNS20I1ztwi4GfREqsas0oYg== X-Complaints-To: abuse@comcast.net X-DMCA-Complaints-To: dmca@comcast.net X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly X-Postfilter: 1.3.32 Xref: dp-news.maxwell.syr.edu comp.databases.theory:34210 "Frank Hamersley" wrote in message news:2eyaf.8082$Hj2.1929@news-server.bigpond.net.au... > VC wrote: >> "Bernard Peek" wrote in message >> news:xGctaeGgBWaDFwgS@shrdlu.com... >> >>>In message , VC >>> writes >>> >>>>>Should these tests really be considered attributes? Wouldn't the act >>>>>of presenting them as a two-tuple relation simple >>>>>be an wise step of normalization - not an example of an EAV >>>>>decomposition? Am I totally missing the boat on this? >>>> >>>>It's a single test with ~6000 measurements. >>> >>>That doesn't sound like something that requires multiple attributes. If >>>it's a single test then the results are presumably in one domain so a >>>single entity should work. >> >> I am not intimately familiar with the testing process. According to the >> person who created the model, it's a drug discovery chemical compound >> testing process which runs daily. The test result for a given compound >> is represented by many thousands of numbers, hence 600 attributes per >> entity. They generated close to half a terabyte of experimental data each >> day. >> >>>What distinguishes one set of test results from another? Is it the time >>>the test was done, the temperature? > > It seems to me that this sounds like a situation where practical concerns > to cope with the volume of data has induced an attempt to achieve > compression (by reducing the repeated occurence of key information were > the schema to be normalised per RM theory. "Compression" or disk space saving was not a consideration. As I said several time before, bad performance due to excessive unions/joins was. They simply could not process the data o time. > If the first model produces 1/2 Tb then the second will surely occupy much > more space. See above. >don't think I could ever sell out the RM and go with option 1. It is much >better IMO to find ways to overcome the problems of the second option by >choosing the right DBMS and configuring it appropriately, rather than (to >coin a metaphor) to "use an SUV (or even a fleet of SUV's) to move 20 >tonnes of freight from coast to coast"! As I said before, replacing SQL Server with another db was not an option due to non-technical reasons.