5 ms·

My reaction to this article: Mwahahaha. Although surprised it is Cassandra, and not mysql thats taking the market share. Is there some technical reason for thi

My reaction to this article: Mwahahaha. Although surprised it is Cassandra, and not mysql thats taking the market share. Is there some technical reason for this? I'm not intimate with what Cassandra offers.

coolsunglasses 11y ago
Probably because of the DataStax OLTP/OLAP/All-Singing/All-Dancing platform that uses Cassandra. MySQL alone doesn't cover the same bases.
StevePerkins 11y ago
Cassandra is a distributed database that easily scales more or less to infinity. As you add more nodes to a cluster, it automatically redistributes data throughout the cluster. It "just works", and that's pretty exciting. So Cassandra at the moment seems to be enjoying a bit of the hype cycle that MongoDB enjoyed 5 or 6 years ago. It's the thing that everyone wants to try out and get onto their resume at the moment. However, like all "NoSQL" databases, the trade-offs come with the usual severe limitations (e.g. no joins, no foreign keys, no transactions in the usual sense of the term, eventual consistency, etc). Curious people diving into Cassandra (or most NoSQL databases for that matter) typically have zero understanding of these trade-offs at first, and think that they will be easier to work with. It isn't, and it isn't supposed to be. It's harder to work with, and has fewer features, because that's the trade-off you're making for scalability. There's little reason to use it if you don't need to scale like that. So, curiosity and RDD ("resume-driven development") aside, I don't think there are many situations in which one would be choosing from Cassandra or MySQL/MariaDB/PostgreSQL. If you are working with multiple terabytes or more of data, then there is seldom a good reason to consider a relational database. If you are working with less than a terabyte of data, then there is seldom a good reason to consider anything OTHER than a relational database. I just don't see much overlap in which these technologies seriously compete against each other.
- acchow 11y ago
  > Cassandra "just works" It's hard for me to believe statements like this. At this point, I think the only things that "just work" are gravity, relativity, and modern CPUs. I recommend this talk on Scaling Pinterest http://www.infoq.com/presentations/Pinterest http://www.infoq.com/presentations/Pinterest
- pmontra 11y ago
  Another trade off is speed vs data duplication. http://blog.websudos.com/2014/08/16/a-series-on-cassandra-part-1-getting-rid-of-the-sql-mentality/ http://blog.websudos.com/2014/08/16/a-series-on-cassandra-pa... "Duplication of data is not only possible, but it’s also encouraged and often the only way to model the SQL equivalent."
mappingbabeljc 11y ago
Hiya. I wrote the article (for verification see Twitter handle @mappingbabel). In poll of 20 $1bn> startups vast majority used MySQL, though I saw a few mentions of Cassandra and MongoDB. Additionally (though not really DBMS) tons of ElasticSearch and Redis. We pull out Cassandra DB specifically because we were able to find a reasonable price comparison example for it (most people I spoke to were just rolling their own MySQL rather than going to a vendor).
davidgerard 11y ago
Postgres is at last eating Oracle from the bottom up, doing what MySQL tried to do fifteen years ago and wasn't quite technically up to. We're busy replacing our Oracle with Postgres. It's just ridiculously more agile. Actually-good databases without having to think about licensing? Heresy!

coolsunglasses 11y ago

Probably because of the DataStax OLTP/OLAP/All-Singing/All-Dancing platform that uses Cassandra. MySQL alone doesn't cover the same bases.

StevePerkins 11y ago

Cassandra is a distributed database that easily scales more or less to infinity. As you add more nodes to a cluster, it automatically redistributes data throughout the cluster. It "just works", and that's pretty exciting. So Cassandra at the moment seems to be enjoying a bit of the hype cycle that MongoDB enjoyed 5 or 6 years ago. It's the thing that everyone wants to try out and get onto their resume at the moment. However, like all "NoSQL" databases, the trade-offs come with the usual severe limitations (e.g. no joins, no foreign keys, no transactions in the usual sense of the term, eventual consistency, etc). Curious people diving into Cassandra (or most NoSQL databases for that matter) typically have zero understanding of these trade-offs at first, and think that they will be easier to work with. It isn't, and it isn't supposed to be. It's harder to work with, and has fewer features, because that's the trade-off you're making for scalability. There's little reason to use it if you don't need to scale like that. So, curiosity and RDD ("resume-driven development") aside, I don't think there are many situations in which one would be choosing from Cassandra or MySQL/MariaDB/PostgreSQL. If you are working with multiple terabytes or more of data, then there is seldom a good reason to consider a relational database. If you are working with less than a terabyte of data, then there is seldom a good reason to consider anything OTHER than a relational database. I just don't see much overlap in which these technologies seriously compete against each other.

acchow 11y ago
> Cassandra "just works" It's hard for me to believe statements like this. At this point, I think the only things that "just work" are gravity, relativity, and modern CPUs. I recommend this talk on Scaling Pinterest http://www.infoq.com/presentations/Pinterest http://www.infoq.com/presentations/Pinterest
pmontra 11y ago
Another trade off is speed vs data duplication. http://blog.websudos.com/2014/08/16/a-series-on-cassandra-part-1-getting-rid-of-the-sql-mentality/ http://blog.websudos.com/2014/08/16/a-series-on-cassandra-pa... "Duplication of data is not only possible, but it’s also encouraged and often the only way to model the SQL equivalent."

mappingbabeljc 11y ago

Hiya. I wrote the article (for verification see Twitter handle @mappingbabel). In poll of 20 $1bn> startups vast majority used MySQL, though I saw a few mentions of Cassandra and MongoDB. Additionally (though not really DBMS) tons of ElasticSearch and Redis. We pull out Cassandra DB specifically because we were able to find a reasonable price comparison example for it (most people I spoke to were just rolling their own MySQL rather than going to a vendor).

davidgerard 11y ago

Postgres is at last eating Oracle from the bottom up, doing what MySQL tried to do fifteen years ago and wasn't quite technically up to. We're busy replacing our Oracle with Postgres. It's just ridiculously more agile. Actually-good databases without having to think about licensing? Heresy!