Topic: How radical ideas take hold (Read 1042 times)
Nosql databases still struggle to take any hold, mysql is the king. And MPP databases are kicking their arses hard.
If you read the full article linked in the References section of my post, you will get more info about the RDBMs topic. In a nutshell, RDMS is good for generating reports and inverting data on its head. The data can be progressively duplicated between an object database, and a relational database, or exported/imported between one and the other on a need basis. It also allows an administrator to inspect the data without writing any algorithmic code.
I observed that since day 1 of these relational databases people wasted an inordinate amount of time writing glue code between sql and the OO-model. It is really sad the amount of man-hours wasted doing that. The OO-RDBM mapping framework are helping to solve that, but we are doing the work of the compiler when using those. It is however, very hard to work efficiently with the database using those models.
The point I am making in my blog post is that what is in fashion changes independently of the actual ideas, but more based on circumstances.
Bitcoin is also an example. We had to wait for the inflation crisis to get tracking with a non-inflationary currency.
I observed that since day 1 of these relational databases people wasted an inordinate amount of time writing glue code between sql and the OO-model. It is really sad the amount of man-hours wasted doing that. The OO-RDBM mapping framework are helping to solve that, but we are doing the work of the compiler when using those. It is however, very hard to work efficiently with the database using those models.
The point I am making in my blog post is that what is in fashion changes independently of the actual ideas, but more based on circumstances.
Bitcoin is also an example. We had to wait for the inflation crisis to get tracking with a non-inflationary currency.
They scale perhaps for internal corporate usage. Not for a large website. MySQL is the best at that. But even so there are limits.
http://gigaom.com/cloud/facebook-trapped-in-mysql-fate-worse-than-death/
Quote
Facebook has split its MySQL database into 4,000 shards in order to handle the site’s massive data volume, and is running 9,000 instances of memcached in order to keep up with the number of transactions the database must serve.
http://gigaom.com/cloud/facebook-trapped-in-mysql-fate-worse-than-death/
Do relational databases like Postgres and Oracle really not scale? I had thought the no sql stuff was just people who liked to do things differently.
Postgres and Oracle scale to the commercial needs of large organizations, and large government departments. But they don't scale to the needs of, say, a search engine. Sure, you could implement Google's index of the WWW using Postgres or Oracle, but not at a reasonable cost because a relational database makes much heavier demands on the hardware.Virtually all NoSQL users would acknowledge that SQL would be preferable, if it weren't for price and performance considerations.
Quote
About ten years ago I stumbled upon an article by a computer programmer Hubert Tonneau, the author of the Pliant programming language. He argued against the relational database:
At that time I looked around to find similar opinions on the internet but found none. Moreover, when I repeated it to my colleagues I was ridiculed: "Nonsense, we abandoned hierarchical database exactly because it wasn't good, and relational one is better". But look around now, we are now in the age of NoSql, Redis, and Mongo databases. How did that happen ?
At that time I looked around to find similar opinions on the internet but found none. Moreover, when I repeated it to my colleagues I was ridiculed: "Nonsense, we abandoned hierarchical database exactly because it wasn't good, and relational one is better". But look around now, we are now in the age of NoSql, Redis, and Mongo databases. How did that happen ?
Ten years ago was the height of LDAP's trendiness. Today it quietly backs most large-scale authentication databases and technologies like Active Directory which hipster admins regularly cream their pants over. You aren't aware of this because old, functional, standardized technologies aren't the stuff of blogs.
How did we get here? Relational databases don't scale. Few organizations wanted to put the resources into supporting hybrids (separating hierarchical data). And it turns out that throwing hardware at a kludge like NoSQL makes more money in the short term.
So, basically, we got here by doing things as half-assed as humanly possible. That's the story of practical computing in general. In fact I could probably turn this into the fault of the Federal Reserve, given enough whitespace. But I'll stop there for now.
Do relational databases like Postgres and Oracle really not scale? I had thought the no sql stuff was just people who liked to do things differently.
Quote
About ten years ago I stumbled upon an article by a computer programmer Hubert Tonneau, the author of the Pliant programming language. He argued against the relational database:
At that time I looked around to find similar opinions on the internet but found none. Moreover, when I repeated it to my colleagues I was ridiculed: "Nonsense, we abandoned hierarchical database exactly because it wasn't good, and relational one is better". But look around now, we are now in the age of NoSql, Redis, and Mongo databases. How did that happen ?
At that time I looked around to find similar opinions on the internet but found none. Moreover, when I repeated it to my colleagues I was ridiculed: "Nonsense, we abandoned hierarchical database exactly because it wasn't good, and relational one is better". But look around now, we are now in the age of NoSql, Redis, and Mongo databases. How did that happen ?
Ten years ago was the height of LDAP's trendiness. Today it quietly backs most large-scale authentication databases and technologies like Active Directory which hipster admins regularly cream their pants over. You aren't aware of this because old, functional, standardized technologies aren't the stuff of blogs.
How did we get here? Relational databases don't scale. Few organizations wanted to put the resources into supporting hybrids (separating hierarchical data). And it turns out that throwing hardware at a kludge like NoSQL makes more money in the short term.
So, basically, we got here by doing things as half-assed as humanly possible. That's the story of practical computing in general. In fact I could probably turn this into the fault of the Federal Reserve, given enough whitespace. But I'll stop there for now.
Jump to: