3 ms·
Drizzle - A Lightweight SQL Database for Cloud and Web
- newt0311 18y agoFrom earlier news: No views, no stored procedures, no prepared statements, trigger, and who knows what else. Seriously, THIS IS NOT A DATABASE.
- azsromej 18y agoIt's well suited to a certain class of applications, specifically web apps. I've never needed the things you list and I'm not the only one.
- newt0311 18y agoI am sure that you are not the only one who requires such an application. Regardless, that is not a database (certainly not a relational/SQL one), it is effectively a dumb object store. I meant nothing derogatory when I pointed out that drizzle was not a database. Just that databases are a very well defined utilities (cf. ANSI SQL standard). MySQL already pushed the bottom end of the envelop. Calling drizzle a database is a flat out lie. Heck, if I ever just want a transactional system for storing undifferentiated blobs, I will probably end up using this product.
- henning 18y agoMore generally, there are basically two kinds of apps that connect to a database: "application databases" that are the primary or sole user of a database and use it for all their persistence needs, and "integration databases" that are one of many interacting with the database. Web apps usually work with application databases. AFAIK Martin Fowler coined this terminology/distinction, see http://martinfowler.com/bliki/ApplicationDatabase.html http://martinfowler.com/bliki/ApplicationDatabase.html . For application databases, usually having any kind of view/stored procs/triggers/prepared statements is a smell that you're conflating business logic with persistence.
- mdasen 18y agoI'm not convinced that this is useful as a webapp data store. What is needed for a webapp datastore? Well, in my opinion, it needs replication and reliability above anything else. Drizzle doesn't offer that. Drizzle offers a MySQL that will scale up a little further sot that you don't have to worry about replication when you have 1M hits per month, but 2M. That's not useful (to me). Partially because at that 2M point, you're in trouble and I don't like that. Mostly because even if I don't need replication for performance, I do need it for availability. I don't mind if a datastore even removes joins. I want highly reliable, highly available, highly redundant, highly scalable data storage. I guess that's why CouchDB is more interesting to me.