Re: [GENERAL] Could use some advice on search architecture

I ended up running some tests using 5 million rows of products. I used about 5 properties that a product should always be matched to, and then I used the following in the select; (CASE property1 in (option1, option2, option3, etc) WHEN TRUE THEN 1 ELSE 0 END) + (CASE property2 in (option1, opt

Re: [GENERAL] Could use some advice on search architecture

the other is named imcs. imcs is mainly a memory database,it is very fast for olap,because of multi threads parallel query plan and column storage.for billions of data,most group by like querys return within 2s. we use it for several weeks and happy for the performance. jov 在 2014-4-19 下午8:27，"Dor

Re: [GENERAL] Could use some advice on search architecture

Because this is a topic that interests me I have done some digging. MonetDB is a proper column-store DBMS. It is used on some decent sized projects, several of which are EU-funded. Digging a bit deeper I have discovered that a PostgreSQL Foreign Data Wrapper for M

Re: [GENERAL] Could use some advice on search architecture

>On 04/19/2014 06:26 AM, Ron Pasch wrote: >> > - It should be possible to search for products and provide properties >> > that the product SHOULD have, not must have. >>> >>> I don't understand this. Say you have a sprocket in red and green. Do you >>> want to search for: >>> >>> select * from

Re: [GENERAL] Could use some advice on search architecture

On 04/19/2014 06:26 AM, Ron Pasch wrote: > - It should be possible to search for products and provide properties > that the product SHOULD have, not must have. I don't understand this. Say you have a sprocket in red and green. Do you want to search for: select * from product where name = '

Re: [GENERAL] Could use some advice on search architecture

Postgresql has 2 column store, 1-in memory(cant remember the name) and http://www.citusdata.com/blog/76-postgresql-columnar-store-for-analytics On Sat, Apr 19, 2014 at 2:10 PM, Robin wrote: > bottom post > On 19/04/2014 12:46, R. Pasch wrote: > > On 19-4-2014 9:38, Robin wrote: > > > Well, giv

Re: [GENERAL] Could use some advice on search architecture

bottom post On 19/04/2014 12:46, R. Pasch wrote: On 19-4-2014 9:38, Robin wrote: Well, given that there are known limited attributes, this is the type of application that really really suits a column oriented database, such as Sybase IQ (now sold by SAP). Its a neat product that scales. Grea

Re: [GENERAL] Could use some advice on search architecture

quot;Robin [robin...@live.co.uk] Gesendet: Samstag, 19. April 2014 09:38 An: pgsql-general@postgresql.org Betreff: Re: [GENERAL] Could use some advice on search architecture On 18/04/2014 21:24, Andy Colson wrote: > On 4/18/2014 8:59 AM, Ron Pasch wrote: >> Hello, >> >> I'

Re: [GENERAL] Could use some advice on search architecture

On 18/04/2014 21:24, Andy Colson wrote: On 4/18/2014 8:59 AM, Ron Pasch wrote: Hello, I'm contemplating what architecture I should use to make searching as fast as possible given the information available and the search requirements. Let me give some background first; - The database contains

Re: [GENERAL] Could use some advice on search architecture

On 4/18/2014 8:59 AM, Ron Pasch wrote: Hello, I'm contemplating what architecture I should use to make searching as fast as possible given the information available and the search requirements. Let me give some background first; - The database contains products of can potentially have a lot of

Re: [GENERAL] Could use some advice on search architecture

First, I would not restrict color to 30 colors, if anything like furniture or clothing, etc. is involved. Colors are very important to consumers, and exact colors are important. I would re-think my color selections. Make sure you have indexes on all the appropriate columns, of course. Susan O

[GENERAL] Could use some advice on search architecture

Hello, I'm contemplating what architecture I should use to make searching as fast as possible given the information available and the search requirements. Let me give some background first; - The database contains products of can potentially have a lot of them (up to about 3 to 5 million) -