Amazon Web Services Announces Relational Database Services

Written by Alex Williams / October 27, 2009 2:43 PM / 5 Comments

Amazon is providing users with the ability to run relational databases in the cloud. The service, Amazon Relational Database Service (RDS), means that customers now have another way to use a cloud service for a function normally administered by an IT department.

Werner Vogels, chief technology officer for Amazon, says the new service means that RDS customers will not have to deal with "the 'muck' of relational database management freeing up its users to focus on their applications and business." RDS will take care of the headaches such as patching and IT administration of the relational database.

According to Vogels, RDS provides the full capabilities of a MySQL Database. The goal is to make it simple for customers to move their databases to RDS without needing to make modifications. Like all Amazon Web Services, RDS scales based on usage. The customer may use as much storage, power and memory as they need. Costs ebb and flow depending on how much they use. Storage, for instance, is managed through API calls.

RDS pricing per instance is as follows:

Amazon Relational Database Service (Amazon RDS).jpg

In addition to RDS, Amazon now offers Simple DB and Amazon EC2- Relational Database AMI.

The Simple DB service is a simpler version of RDS. According to Vogels, it is designed for applications that do not require a relational model and principally demand indexing and querying capabilities.

The Amazon EC2- Relational Database AMI is designed for users who have particular relational database requirements and want full control.

RDS may be a bit slow to adoption as security is the big issue for the enterprise. But this issue will wane as people recognize that cloud offerings can be as secure as on-premise systems. Further, we are sure that security concerns will mellow when companies compare the costs and time requirements of RDS to what they do on-premise.

In that regard, prices are already decreasing for EC2. Starting November 1, Amazon is dropping its per-hour prices for the following:

Amazon Web Services Blog.jpg

That's a 15 percent drop in cost from the current prices for Linux instances.

Comments

Subscribe to comments for this post OR Subscribe to comments for all ReadWriteEnterprise posts

Well it is interesting how it looks like in the case when someone has already been using a database and would like to migrate to Amazon service for example his Oracle database.

Posted by: Gogi | October 28, 2009 12:48 AM

My company provides file hosting and collaboration ( www.binfire.com ). We are moving our services from our own dedicated servers to Amazon's. The price cut is good news for us. Also the RDS service will help us to move to scalable database. I hope the migration be easy and straight forward! I am not still sure to the extend of code changes we have to do.

Posted by: david Robins | October 28, 2009 1:52 AM

I spent all of last night working with RDS for some massive data sets. It was so disappointing. Setup - excellent. API interface - very good. Query speed - Awesome. Insert, update, delete speeds - HORRIBLE.

Using an "Extra Large EC2 Standard Instance" to process data and a "Quadruple Extra Large DB Instance" for my database, I could only achieve 85 Queries Per Second.

When I setup a MySQL server on my "Extra Large EC2 Standard Instance", I was able to achieve 190 Queries Per Second using the exact same script.

However, even the 190 QPS pales in comparison to a much lower powered Dell 1950 with only 8GB of RAM acting as the DB server and the application server. Again, with the same script and data, I could achieve > 7500 QPS.

Amazon has put some amazing features into the RDS service. The instant snapshots, point in time recovery, etc are great if you have small databases with very little activity. However, if you have any real data processing to do, EC2s and RDS are pretty useless.

Posted by: Justin Noel | October 28, 2009 9:15 PM

Justin, are you writing the apps that are accessing the DB? If so, have you considered partitioning ("sharding") your DB into multiple smaller ones? As I'm sure you know, if you have huge amounts of data (my guess is you do, since you are using Amazon's "Quadruple Extra Large DB Instance"), performance can really crawl. Was your DB the same size on your Dell?

This page discusses sharding and how to implement it in your code: http://bit.ly/2NRKWY

It would be interesting to compare performance numbers between Amazon's RDBMS and Microsoft's SQL Azure (http://bit.ly/SQLAzureDB).

Posted by: Jason (working for M80, representing Microsoft) | October 30, 2009 4:46 PM