Aurora: Soon going to be available for Amazon RDS

Relational Database Service (RDS) is a service by Amazon that manages different kinds of database engines to handle the tasks like provisioning, backups, patching, recovery and failure detection. Amazon has launched a fifth database engine for RDS called Aurora which delivers high performance and is cost effective when compared to current commercial database engines.

The features that make Aurora different from the other commercial relational database engines are:

Optimal design:
All the existing database engines are based on monolithic design model, are complicated and are expensive to maintain. Amazon Aurora has overcome these limitations with a new design after years of development lifecycle.
In the traditional database architectures, functionalities like transactions, caching and logging are all performed and reside within a single layer. In order to scale this kind of an architecture, the entire stack has to be replicated. So, leveraging the benefits of cache, logs and storage cannot be achieved.
The architecture of Aurora is designed using a service oriented approach and the the logging and storage layers are loosely coupled and can be scaled out with flexibility.

MySQL compatible:
Aurora is fully compatible with MySQL 5.6. So the existing code, drivers and applications can be used with a little or no change.

Performance:
On Aurora one can perform one lakh inserts per second and five lakh selects per second which is 5 times faster than MySQL running on high end servers.

Scalable Storage:
Aurora is backed with SSD scaleout multi tenant storage which automatically scales out as the database grows. Storage can be scaled up to 64TB by adding 10 GB chunks as per need without affecting the database performance. So with Aurora, there is need of worrying about provisioning the storage or IOPS.

Log Structured Storage System:
Aurora includes a new log structured storage system that is scalable, multi-tenant and optimised for DB workloads. In this architecture, redo logs are spread across storage segments that is each segment has its own redo logs.

Highly Durable:
Using Aurora you can replicate your data in six ways across three availability zones. Read quorums are set to the value of three out of six and write quorums are set to the value of four out of six. In this model, the performance of database is not affected even if two nodes are down. Data is automatically, incrementally and continuously backed up to S3 which is designed for 99.999999999% durability.

Highly Available:
It is designed to handle a loss of more than one copy of data without affecting the database performance. Aurora continuously scans and repairs disk errors.

Quick Crash Recovery:
In Aurora the buffer cache is separated from the database layer. So, the cache is always available even in the case of events like reboot or crash. Making cache persistent during the reboot will reduce the wait time of buffer cache to get populated which otherwise may take hours for other DB engines to accomplish.

Cost-effective:
Aurora is priced at one-tenth of the cost of the most popular database engines and delivers the same performance and availability like them.

For now the Aurora is available as preview only in the North Virginia region. The security features like encryption are not introduced in preview. Hopefully Amazon will come up with better encryption policies by the time of release of this service.

WRITTEN BY CloudThat