Jim Lawrence
accessd at shaw.ca
Tue Feb 5 12:01:03 CST 2013
Hi All: I worked on a huge database that held all the government's and in particularly the Attorney Generals Ministry financial records. The one giant data storage held all the account receivable, purchase orders, client ledgers, accounts payable, general ledgers and so on. It was all in one giant distributive data storage system and it ran on B-tree structure of keys and indexes. The system was running on equipment that we would think primitive by today's standards but from any terminal, in the province, a clerk, depending on clearance rights, could access any records or grouping in seconds. The most active users were connected through coax cables; bundles and bundles of them. (actually did some of the wiring...definitely not a pleasant task) Most of the remote terminal connections were connected via a 300 baud modems, the data source was in one room with six UNIX servers (eventually Linux boxes). This whole system was not relational as we would know it but it did the job very effectively. The old database has been moved to an Oracle database system and the costs were about 100 millions dollars to do it. The new system requires over a thousand servers. It now even requires overnight batches to retrieve a special reports and according to the users the whole system run significantly slower. This is one clear example where relational databases just do not cut it. I am sure there are many examples where traditional databases just don't work. Much of the internet runs on non-relational databases. With most of our clients, they just don't have enough data to worry about and we could get away with series of cascading hierarchical spreadsheets that could furfill all their needs. Relational databases with a SQL FE are just the standards we use to but are hardly the only solutions. Jim -----Original Message----- From: dba-tech-bounces at databaseadvisors.com [mailto:dba-tech-bounces at databaseadvisors.com] On Behalf Of Stuart McLachlan Sent: Tuesday, February 05, 2013 3:01 AM To: Discussion of Hardware and Software issues Subject: Re: [dba-Tech]“Depend on me, and I’ll set you free!” Rant is the word all right. There are so many things wrong with his arguments that I don't know where to begin. But a few immediate points: He clearly can't tell the difference between "relational database" and "SQL". He thinks that a database engine which uses BTree indexing can't be "relational" (He obviously never worked with Dataflex) His characterisation of all databases as "big, stogy, slow, expensive pain in the rear." demonstrates a clear lack of experience in using them. How the h*ll can you develop and *test* the use cases and business rules without the underlying data? The data model IS the core of any real world bsiness system. It's tragic that "an award winning author, renowned speaker, and über software geek since 1970" has carried the same baggage around with him for 40 odd years. -- Stuart P.S. He was right about one thing though - American beer. Until recently it was like sex in a canoe. On 5 Feb 2013 at 13:52, Salakhetdinov Shamil wrote: > Hi All -- > > It's funny I've just planned to try to implement one experimental > project using "No DB" (NoSQL) approach - and here I have got an > interesting "rant article" arrived :) > > http://blog.8thlight.com/uncle-bob/2012/05/15/NODB.html > > Thank you. > > -- Shamil > _______________________________________________ > dba-Tech mailing list > dba-Tech at databaseadvisors.com > http://databaseadvisors.com/mailman/listinfo/dba-tech > Website: http://www.databaseadvisors.com _______________________________________________ dba-Tech mailing list dba-Tech at databaseadvisors.com http://databaseadvisors.com/mailman/listinfo/dba-tech Website: http://www.databaseadvisors.com