The point in NoSQL is mostly in group (3), web2+ startups with a possible 1B user and 1000B things (comments etc) per user. Joins and normalization here are a bit costlier and what is written in your Oracle development manuals, for your small workgroup intracompany app don't work here.
So the group (0), which are mostly db developers of the client/server architecture; when they attack the web2.0 problems, they fail because they stick to dogmatic notions as if they are true. Though (2) people are ignorant of relational concepts, (0) people are stubborn, uneducatable people who end up creating all types of scalibility problems. They overuse the notions of normalization, but forget they attack the wrong problem with the wrong tools.
PS: I am not yet using NoSQL in production and have a solid past in Oracle/Db2/Ms SQL/Sybase, and now now doing startups in MySQL/Postgres and Mongo.
I have to disagree with you. Normalisation is a very basic concept. Whilst it can cause a couple of problems, I am very sceptical about most startups hitting them. I'd like to see DBs introduce denormalisation as a feature, separating the logic from practicality.
Having said this, I like working with MongoDB. I like schemaless design and flexibility.
Comments
The point in NoSQL is mostly in group (3), web2+ startups with a possible 1B user and 1000B things (comments etc) per user. Joins and normalization here are a bit costlier and what is written in your Oracle development manuals, for your small workgroup intracompany app don't work here.
So the group (0), which are mostly db developers of the client/server architecture; when they attack the web2.0 problems, they fail because they stick to dogmatic notions as if they are true. Though (2) people are ignorant of relational concepts, (0) people are stubborn, uneducatable people who end up creating all types of scalibility problems. They overuse the notions of normalization, but forget they attack the wrong problem with the wrong tools.
PS: I am not yet using NoSQL in production and have a solid past in Oracle/Db2/Ms SQL/Sybase, and now now doing startups in MySQL/Postgres and Mongo.
Your group (3) is just his group (1) stated differently.
He denormalized the data.
yea, I show the nosql way by repeating the data.
I have to disagree with you. Normalisation is a very basic concept. Whilst it can cause a couple of problems, I am very sceptical about most startups hitting them. I'd like to see DBs introduce denormalisation as a feature, separating the logic from practicality.
Having said this, I like working with MongoDB. I like schemaless design and flexibility.