Amazon announces RDS Read Replicas
This piece of news went pretty much unnoticed to the wider community: Amazon RDS Announces Read Replicas.
SimpleDB is simply not enough for most enterprise users. This is why Amazon, last year, released its Amazon Relational Database Service. Anybody who managed a high-traffic site knows that managing a database server is one of the most painful duties: databases need to be redundant, secure, non-corrupted, and -- most importantly -- need to be backed up. When your DB very big, backups become tricky at best. You end up spending incredible amounts of resources just making sure your database servers work correctly -- and, every now and then, going insane getting things to work again after a disaster. Some companies never, ever recover from a database disaster.
Before now, you could have Multi-AZ deployments of your database server, but you could only access one database at a time -- the replicas were always in "stand by", and weren't available for queries. The main purpose of Multi-AZ deployments is having replicas in case the main database server is no longer available.
With Read Replicas, you can have several databases that can be used as read-only replicas of the main one (which is the only one you can write on). Unlike Multi-AZ deployments, Read Replicas can fall behind the main database: this is why they shouldn't be used in case the main database isn't available.
As far as we are aware, this step is the last piece of the puzzle in terms of Amazon's offering. The lack of Read Replicas was a huge withdraw for people with highly trafficked sites (which, ironically, was the main target of Amazon's S3 services). With this step, people with limited specialised knowledge of database architectures can create a very strong systems with Multi-AZ deployments (where different nodes take over if the main one goes down), and with several Read Replicas serving read queries.
Kudos to Amazon for this great advancement.
