About PayScale Human Capital
Launched in January 2002 and headquartered in Seattle, PayScale has been giving light to a once dark area by providing salary, benefits, and compensation information. The service not only helps job candidates be informed about what they are worth, but also helps companies make competitive salary offers. PayScale has collected over 50 million individual salary profiles to create the world’s largest salary information database. The service works by enabling individuals to submit their job and salary profiles and be compared to other individuals in the market. With this volume of information, the company has been able to statistically induce and cater real-time salary information.
PayScale hosts several large Microsoft SQL Server databases with high I/O workload. It is imperative to have measures in place to protect data and be able to recover from any failure without compromising performance of production databases. The company wanted to update their Business Continuity and Disaster Recovery Plan for the databases in the event of a full data center outage and needed a backup site with the infrastructure that would not only keep up with demand and easily scale but also offered security. This is where Amazon Web Services (AWS) was the right choice.
The AWS Solution
Due to the workload of the databases and other existing technologies in place, it was determined using native SQL Server Database Replication would provide the least intrusive way of copying data to the DR site. Amazon EC2 became the natural choice for the DR server. Although Amazon RDS service is available, having a Microsoft SQL Server on EC2 allowed the deeper level of database administration needed to setup this particular replication.
In terms of security, the Microsoft SQL Server DR instance is placed inside a VPC. Amazon Virtual Private Cloud (VPC) is a logically isolated virtual network in AWS in which instances and other resources can be securely provisioned. The VPC service allowed staff to connect to the Microsoft SQL Server DR instance via their existing VPN while protecting the instance from unauthorized access.
Data replicates asynchronously from the datacenter to AWS over VPN connection. A staging instance was setup to be the publisher for the DR databases. The figure below shows the resulting database DR architecture.
Figure 1: PayScale database DR architecture
The Success Story
Aside from providing a DR site for the databases, AWS also became a playground for doing database development and load testing. It became easy to launch a test environment by launching a SQL Server instance with the EBS snapshot of the DR instance. PayScale has also realized upfront cost savings by using AWS Reserved Instance as opposed to infrastructure on hardware. AWS has helped the company’s agility with cost savings and business continuity.
Cascadeo continues to support and provide MS SQL advice to PayScale to keep the systems running.