forrest@blia.BLI.COM (Jon Forrest) (05/22/87)
There have been several postings recently regarding benchmarks of the popular RDBMS's. I'd like to remind people that test benchmarks are well and good but the performance of the production system often falls short of the expectations gained from tests. One way to help remedy this problem is to use a database machine. Of course, the same difference between test and production performance could occur on a database machine, but the performance of a database machine will probably let you do more before this happens. This fact has helped us sell our database machines to places that have already bought a software-only RDBMS only to find that when push came to shove, what they had wasn't fast enough. Our new database machine, which I'm told will be 10 times faster than what we already have, will make this even more relevent. Of course, this will also be true with our competitors products. Jon Forrest ucbvax!mtxinu!blia!forrest
jeff@rtech.UUCP (Jeff Lichtman) (05/24/87)
in article <2700@blia.BLI.COM>, forrest@blia.BLI.COM (Jon Forrest) says: > > One way to help remedy this problem is to use a database machine... > > This fact has helped us sell our database machines to places > that have already bought a software-only RDBMS only to find that > when push came to shove, what they had wasn't fast enough... > > Jon Forrest I object to this type of posting. There have been many times in the past when I have been tempted to plug Relational Technology in this newsgroup, but I have always resisted. To do so would have been inappropriate. I thought there was a general rule that vendors are not supposed to use the net to promote their products. For example, you don't see chip manufacturers filling up comp.arch with arguments about why their CPUs are better. I believe that DBMS vendors' contributions to comp.databases should be purely technical and as objective as possible. Without this, the content of the newsgroup will devolve into propaganda. -- Jeff Lichtman at rtech (Relational Technology, Inc.) "Saints should always be judged guilty until they are proved innocent..." {amdahl, sun}!rtech!jeff {ucbvax, decvax}!mtxinu!rtech!j unC
forrest@blia.BLI.COM (Jon Forrest) (05/25/87)
In article <851@rtech.UUCP>, jeff@rtech.UUCP (Jeff Lichtman) writes: > in article <2700@blia.BLI.COM>, forrest@blia.BLI.COM (Jon Forrest) says: > > > > One way to help remedy this problem is to use a database machine... > > > I object to this type of posting. There have been many times in the past > when I have been tempted to plug Relational Technology in this newsgroup, > but I have always resisted. To do so would have been inappropriate. I don't agree. If there were some reason you thought your product was better than the competition for technical reasons then I think a posting would have been in order, especially if you prefaced your posting title with the fact that it contained a plug for your company. I recently complained that the ba.seminars was too commercial because seminars were being announced that had fees that were for more than the cost of putting on the seminar. Much to my surprise, the responses to my complaint were not in agreement with me. Apparently, people's threshold of sensitivity toward commercialness isn't has low as mine. > I thought there was a general rule that vendors are not supposed to use the > net to promote their products. For example, you don't see chip manufacturers > filling up comp.arch with arguments about why their CPUs are better. I believe Oh yes you do. The point is that the discussion is technical. If you re-read my posting you will notice that it is actually discussing one technical advantage of database machines over software only RDBMS's. I didn't mention any aspects of our database machine. In fact, since we now have some competition (Teradata, Sybase) my posting is hardly a plug for Britton Lee. Also notice that I didn't single out any one software RDBMS. So, if any of you think that your company's product is better than the competition I, for one, would like to hear about it. Let's just try to keep the discussion technical. Jon Forrest ucbvax!mtxinu!blia!forrest
authorplaceholder@tiger.UUCP.UUCP (05/26/87)
Bravo Jeff!!! Discussions of technology are wonderful and hopeful, but if I want an advertisement, I'll read a tabloid. In this newsgroup, at least, I for one appreciate the discretion of the DBMS vendors (excepting one of course) to not blatantly plug their respective products. After all, performance is relative to the user's application needs anyway. Jeff McReynolds AT&T Network Systems Oklahoma City, OK ...ihnp4!occrsh!jlm
anderson@aero.UUCP (05/26/87)
In article <851@rtech.UUCP> jeff@rtech.UUCP (Jeff Lichtman) writes: >in article <2700@blia.BLI.COM>, forrest@blia.BLI.COM (Jon Forrest) says: >> >> One way to help remedy this problem is to use a database machine... >> >> This fact has helped us sell our database machines to places >> that have already bought a software-only RDBMS only to find that >> when push came to shove, what they had wasn't fast enough... >> >> Jon Forrest > >I object to this type of posting. > .... >-- >Jeff Lichtman at rtech (Relational Technology, Inc.) I too object, especially to the blattenly misleading advertizement. I use several Britton Lee's in different os environments; I also use RTI's INGRES and Oracle's ORACLE, in addition to another dozen products. AND I REFRAIN ALWAYS FROM PLUGGING ONE OVER THE OTHER IN GENERAL DISCUSSION!! I ESPECIALLY NEVER COMPARE APPLES AND ORANGES - H/W DBMS AND S/W DBMS. I can speak clearly on the similarities and differences between features, and can provide reasonable measures of performance for specific configurations. BETTER is a relative that only makes sense for well described environments. I agree with Jeff that this forum is for discussion of database technologies. I wonder why it is instead used for requesting ridiculous survey results, with even less substantial advertisements in response.
ecec@ur-tut.UUCP (Eric Carleen) (05/27/87)
I'd be interested in hearing plugs by vendors for their own products.
I hope the plugs would be technical. It would be a great way to learn
about the various databases; certainly no one is as familiar with the
shortcomings of database X than the maker of database Y.
If ridiculous, false, or unfair claims were made, they would be subject
to the criticism of the readers of this group, which includes the other
vendors. The result would be an informed discussion greatly superior to
the advertisments that are in the trade publications.
I'm aware that no one wants to see this newsgroup turned into one long
commercial, but I'd also like to learn more about the products that are
available.
What do people think?
Eric Carleen
University of Rochester Medical Center
UUCP: {seismo|allegra}!rochester!ur-msbvax!edc
Bitnet: heartedc@uorhbv
Phone: (716)-275-5391itkin@bene.UUCP (Steven List) (05/28/87)
In article <11872@aero.ARPA> anderson@aero.UUCP (Tom Anderson) writes: >In article <851@rtech.UUCP> jeff@rtech.UUCP (Jeff Lichtman) writes: >>in article <2700@blia.BLI.COM>, forrest@blia.BLI.COM (Jon Forrest) says: >>> One way to help remedy this problem is to use a database machine... >>> >>> This fact has helped us sell our database machines to places... >>> >>> Jon Forrest >> >>I object to this type of posting... >>-- >>Jeff Lichtman at rtech (Relational Technology, Inc.) >I too object, especially to the blattenly misleading advertizement. >I use several Britton Lee's in different os environments; I also use >RTI's INGRES and Oracle's ORACLE, in addition to another dozen products. >AND I REFRAIN ALWAYS FROM PLUGGING ONE OVER THE OTHER IN GENERAL DISCUSSION!! >I ESPECIALLY NEVER COMPARE APPLES AND ORANGES - H/W DBMS AND S/W DBMS. Perhaps I'm in the minority, but I don`t mind at all having employees of various vendors discuss the relative merits of their systems, as well as having users of those systems discuss them. I would be very interested in Tom's evaluation of the differences (note DIFFERENCES not necessarily better or worse) between various systems in different environments and for different applications. We at Benetics have been working with UNIFY for the past five years, and a number of us (myself included) are extremely knowledgeable about UNIFY and ACCELL both from the internals and user perspective. Why should I not share my impressions and opinions here? Isn't this a public forum for discussion? I don't find Jon`s original posting offensive, nor do I object to Jeff and Tom's responses. But I think this is an opportunity for some interesting and educational exchanges. So, Tom, please do post your opinions and evaluations. You don't need to plug to express an opinion. And if you find one to be better than the other in certain circumstances, that may very well benefit me enormously. The same goes for Jeff and Jon - go ahead and express yourselves. I'm anxious to hear what you have to say. I might even express an opinion or two myself somewhere along the way. Steven
peter@utah-gr.UUCP (05/28/87)
I don't think this article was out of line with respect to the "no advertising" understanding (rules wrt usenet postings? you have got to be kidding). It did contain some factual info, and contained a line of reasoning. The point could have been made stronger (my opinion, see disclaimer below) but more importantly it was made in general terms, and did not isolate Britton Lee as THE solution, just as AN example solution. I don't see that it is different from somebody (whether related or not to X or Y) claiming system X does 14 phony baloney transactions per second, while system Y does 12. My view on usenet is that it is pamphleteering in the high tech age. If somebody gets too far out of line wrt advertising or blithering they will be cut off, some how, some way. I think some advertising should be allowed; everyone would like to hear about new features/performance figures/etc even if it is from the horses mouth. As long as the data is "fair", and all parties can be heard, things should work out fine. I guess we could have advertising rot 13'd. Anyways, keep those data base wars going. Disclaimer: I used to work for Britton Lee (before Mr. Forrest's time), have friends that work at Sybase, and think the Teradata machine is pretty neat (and strange). Database machines seem like a good idea to me, what do the people squacking about advertising think about them? Shouldn't the articles on advertising in comp.databases have been posted in news.discussions.advertising(just wondering, this renaming has got me confused)?
dennisg@pwcs.StPaul.GOV (Dennis Grittner) (05/29/87)
In article <1392@ur-tut.UUCP> ecec@tut.cc.rochester.edu.UUCP (Eric Carleen) writes: >I'd be interested in hearing plugs by vendors for their own products. >I hope the plugs would be technical. It would be a great way to learn >about the various databases; certainly no one is as familiar with the >shortcomings of database X than the maker of database Y. As a user of a Britton-Lee and a user of Informix as well as Informix-SQL as well as a former user of Unify I would like to add my thoughts that I'd love to hear from various database ( software and hardware ) folk with TECHNICAL plugs for their product(s). Let's hear debate about the performance per dollar ( how about that one Ingress ) the various methods of networking databases ( Oracle , Ingress how about it ), or about the actual communications interface ( how about that XNS stuff Britton-Lee ) and a whole lot more. I don't want to see a whole lot of drivel from vendors - but it might be darned useful to hear about new concepts and features from the vendors here where we could poke around at them collectively rather than isolating us one at a time at a show or a sales call. "Building nuclear arms for peace is like f***ing for chastity" -- Dennis Grittner City of Saint Paul, Minnesota (612) 298-4402 Room 700, 25 W. 4th St. 55102
walker@cod.UUCP (06/03/87)
Newsgroups: comp.databases Subject: Re: Benchmarks and Real Performance (minor plug for my company) summary: Expires: References: <2700@blia.BLI.COM> <851@rtech.UUCP> <11872@aero.ARPA> <1392@ur-tut.UUCP> <889@pwcs.StPaul.GOV> Sender: Reply-To: walker@cod.nosc.mil.UUCP (Janet M. Walker) Followup-To: Distribution: Organization: Naval Ocean Systems Center, San Diego Keywords: In article <332@bty.UUCP> yost@bty.UUCP writes: >> Perhaps I'm in the minority, but I don`t mind at all having employees of >> various vendors discuss the relative merits of their systems, as well as >> having users of those systems discuss them. Well then, count me in this "minority" too. I, also, am currently evaluating DBMSs (RDBMSs to be exact) and have not ruled out the idea of a "back-end" DB machine to boot :). I have learned some interesting things from this newgroup but would like to learn a lot more that might help me in my decision. We're doing some rapid prototyping (hence RDBMS) for both hardware and software on UNIX based "workstations". Our task entails such things as message processing (involving field checking and storage of "formatted" but variable length text), geographic displays (bit map storage would be nice), sophisticated MMI, and manipulation of more traditional kinds of data (which lends itself well to "traditional" RDBMSs). On top of all this we are going to have a number of "workstations" with several "operators" at each that will be tied together and all have to access the same database in a time-critical fashion. We would like sophisticated 4GL capability AND GOOD PERFORMANCE!! So far I haven't found the "best of all worlds". There are RDBMS products which have capabilities like binary and unlimited text data types but slower performance, etc. I would love to hear from vendors and users alike. Remember too that vendors who tout their products in this forum are subject to criticisms, complaints, and exceptions - what better way to compare sales hype with reality. So let's open up the floor for all! Janet M. Maclaughlin, Code 722 MILNET/ARPANET: walker@nosc.arpa Naval Ocean Systems Center UUCP: [ihnp4,akgua,decvax,dcdwest,ucbvax]! San Diego, California 92152 || (619) 225-2316 (AV) 933-2316 sdcsvax!noscvax!walker
martin@uhccux.UUCP (06/03/87)
In article <711@cod.UUCP> walker@cod.nosc.mil.UUCP (Janet M. Walker) writes: >Keywords: > >In article <332@bty.UUCP> yost@bty.UUCP writes: >>> Perhaps I'm in the minority, but I don`t mind at all having employees of >>> various vendors discuss the relative merits of their systems, as well as >>> having users of those systems discuss them. > >Well then, count me in this "minority" too. I, also, am currently >evaluating DBMSs (RDBMSs to be exact) and have not ruled out the idea of a >"back-end" DB machine to boot :). I have learned some interesting things >from this newgroup but would like to learn a lot more that might help me in >my decision. > Same here. We're evaluating RDMSs for use in distributed medical information systems, running on workstations with multiple fileservers separated in some cases by over a 100 miles. Our requirements include storage of bit-mapped images (1024 x 1024 x 16), e.g., radiographs, MRI and CT scans, sonograms, etc.; support of multiple security class definitions (physician, nurse, patient's physician, emergency room physician, chief radiologist, location, etc.) which allows data to be assigned to multiple security classes according to node/relation/virtual relation (compiled)/field etc. in order to regulate who can access what data in a highly confidentail medical environment; support for data storage on WORM drives (most medical information is by law write-once read mostly). On the mundane financial side, we need heterogeneous database access, since the market we're going after is mired in conventional technology, and we need the ability to interface with IBM databases running on IBM mainframes. Because we are using graphics workstations, our application also requires that the RDBMS have a robust 4GL with a user interface built on top of X-windows, preferably using a user-interface development tool such as Apollo's "Dialog", since X-windows has become a standard for graphics workstations and rumor has it that "Dialog" may become a user-interface standard for graphics workstations. We're also implementing a knowledge-based system to interact with the user entering information into the RDBMS to enforce data integrity from the standpoint of the medical knowledge base, where most of the medical knowledge base is actually stored on the RDBMS. I have a paper which will be published in an upcoming issue of the Journal of Medical Informatics which describes the automated construction of a computer thesaurus type knowledge base using an RDBMS instead of the conventional FRL approach. And we're really not interested in PC implementations, since the low-end Apollo and Sun workstations are now in the same price range as the high-end PCs. We'll be evaluating Ingres, as soon as our demo copy arrives, and have not ruled-out the use of a database machine as opposed to a file/process server running a software RDBMS. Any takers? Regards, Brian K. Martin, M.D. CEO, Martin Information Systems, Ltd. 3420-A Hinahina Street Honolulu, Hawaii 96816 (808) 735-5661 ARPA: uhccux!medix!martin@nosc.mil UUCP: { ihnp4,ucbvax,dcdwest,seismo }!sdcsvax!nosc!uhccux!medix!martin
bert@infoswx.UUCP (06/04/87)
I also think there should be more practical information in this newsgroup. Technical discussions of various methods of accomplishing a task, even if by people who have ties to a commercial interest in a product, can be valuable as long as they don't degenerate into sales blurb hype. The net seems to be fairly good at policing itself in that regard... As for 4GLs - I recently undertook to use Informix-4GL for the development of an applications program here. We use the Informix ESQL, ISQL and C-ISAM products with good results. However, since the application in question involves more than just pure user-interface to and reports from our db, Informix-4GL is not a good tool. The interface to C included is more trouble than it is worth, and restriction after restriction awaits those who try to do anything outside the 4GL, which is required in our case. The 4GL itself is OK, as long as you can do your ENTIRE application in it. This is of course just an opinion I arrived at after working with it for about a month. I would be interested in a posting summarizing your responses to your 4GL question. Bert Campbell @infoswx
bouwhof@cs.vu.nl (Bouhof W) (06/05/87)
To: ecec@tut.cc.rochester.edu.UUCP From: Bouhof W <bouwhof@cs.vu.nl> Subject: Re: Benchmarks and Real Performance (minor plug for my company) Newsgroups: comp.databases In-Reply-To: <1392@ur-tut.UUCP> References: <2700@blia.BLI.COM> <851@rtech.UUCP> <11872@aero.ARPA> Organization: VU Informatica, Amsterdam Cc: Bcc: In article <1392@ur-tut.UUCP> you write: >I'd be interested in hearing plugs by vendors for their own products. >I hope the plugs would be technical. It would be a great way to learn >about the various databases; certainly no one is as familiar with the >shortcomings of database X than the maker of database Y. > >If ridiculous, false, or unfair claims were made, they would be subject >to the criticism of the readers of this group, which includes the other >vendors. The result would be an informed discussion greatly superior to >the advertisments that are in the trade publications. > >I'm aware that no one wants to see this newsgroup turned into one long >commercial, but I'd also like to learn more about the products that are >available. > >What do people think? > > Eric Carleen > University of Rochester Medical Center > UUCP: {seismo|allegra}!rochester!ur-msbvax!edc > Bitnet: heartedc@uorhbv > Phone: (716)-275-5391 Although I'm an enthousiastic UNIX user, I recently also became rather fond of TANDEM's stuff. Here, we are discussing DBMSs and therefore ther should also be a place for Tandem's NonStop SQL (I'm no shareholder) which is called the most relational approach to DBMS and which in a recently performed benchmark (done by Codd and Date's bureau) managed to handle 208 OLTP transactions per second in a simulated 25,000 automatic teller machine environment. In comment to Eric Carleen's article I wish to add that it is indeed interesting to learn more about the available products and to comment (here only briefly) on their merits. Is there anyone who can elaborate on the new NonStop SQL in relation to other (R)DBMS and especially to the myths of the supposed inherent slowness of the relational approach. Is there here a real milestone as stated by Date?? Edwin F. Setzpfand (on vu44!bouwhof)
bh@ptsfa.UUCP (Brian Holliday) (06/10/87)
=As for 4GLs - I recently undertook to use Informix-4GL for the =development of an applications program here. We use the Informix =ESQL, ISQL and C-ISAM products with good results. However, =since the application in question involves more than just =pure user-interface to and reports from our db, Informix-4GL =is not a good tool. The interface to C included is more trouble =than it is worth, and restriction after restriction awaits =those who try to do anything outside the 4GL, which is required =in our case. = =The 4GL itself is OK, as long as you can do your ENTIRE application =in it. = =This is of course just an opinion I arrived at after working with =it for about a month. = =Bert Campbell @infoswx I am sold on Informix 4GL. The interface to C gives me no problems. There are no restrictions that I am aware of. You can integrate C routines into your application, and pass your arguments on the stack (4GL ==> C, and C ==> 4GL). You can also execute a program, by using Informix 4GL's equivalent to "system" (EXECUTE). I would not say that it is perfect, though. You do have to live within the boundaries of the 4GL, as the user interface is in a certain format. There are some bugs (which can be worked around). It is not the fastest database in the world. Compiled 4GL executable binaries are *huge*, which could be a problem if you do not have enough main memory -- the EXECUTE stops working (a 3B5 UNIX problem: the fork hangs) -- but there is a funky work-around. The application I am developing has 2 integrated C routines (one gets shell variables, one is my funky "system" call). I could add more integrated C routines if the application needed it. The application also executes several shell scripts, which execute totally outside of the 4GL environment. One of the system calls gives the user a Bourne shell. I am quite sure that 4GLs are the wave of the future for database programming. There may be better 4GLs than Informix (Informix 4GL is the only one I know, so I can't say), but for my purposes right now, Informix 4GL is good enough. I can develop applications very quickly with it, so it has paid for itself. It does take longer than a month to find all the potholes, and figure out how to work around them, though. NOTE: I do not work for Informix, Inc. These comments are my own, and do not necessarily represent my company's view. Brian Holliday (ptsfa!bh)