Here’s Why AWS Cloud Matters For Life Sciences
[Today’s guest post come from Stu Robertson, head of engineering at the digital health company Foundry Health, and Brock Heinz, Foundry’s CEO.]
Amazon Web Services (AWS) Cloud platform should be of interest to all modern clinical organizations. What is the AWS cloud? It is the shining star of a modern breed of Platform as a Service (PAAS) providers, offering up full datacenter services on demand to run the applications vital to your clinical development.
Why should the life sciences community care? Because leveraging cloud platforms can increase performance, quality and speed to delivery of strategic applications, all while lowering IT costs.
Two key aspects of the AWS platform really make it special.
First is the depth of their infrastructure offerings. We’ll touch on highlights below, but to summarize, an AWS datacenter offers capabilities that few corporate data centers can match. All of this is available on demand and with a 99.95% availability SLA. Leveraging a second data center location in one deployment gives an impressive 99.9975% SLA, and you can deploy as many as five locations for truly astronomical availability SLAs. This is one reason so many major life sciences customers rely on AWS.
Second is the truly geographic reach of AWS. Clinical research is a global enterprise. And while “the cloud” is a term of abstraction, any cloud is composed of physical data centers in real locations. The AWS platform is comprised of many physical clusters of data centers all over the world. What this means in practice is that AWS allows your company’s application portfolio to run geographically close to the clinical operations themselves and have fully redundant instances in another continent if required.
Let’s look at some specific aspects of AWS that led Foundry Health to choose AWS for hosting mission-critical eSource deployments of our flagship product ClinSpark.
AWS takes GxP very seriously. Proactive compliance, external auditors and other protocols are all written up in a thorough overview of GxP suitability. This includes security and quality certifications such asISO 9001, ISO 27001, ISO 27017, ISO 27018,NIST 800-53 under FedRAMP.
Nearly infinite room for vertical and horizontal scaling
With AWS, the size of a server is a configuration. It can be easily and quickly changed. On the menu is a large buffet ranging from small and medium sized servers, all the way up to ones with hundreds of processors and thousands of gigabytes of memory. Increasing the size of a physical server is “vertical scaling,” since the boxes get bigger and taller.
But a bigger box is not always the best way to ensure high performance at high load. Modern web applications typically run on clusters of redundant smaller servers, and the size of this pool can grow and shrink during busy and light times. This is known as horizontal scaling; you can see why below because the diagram becomes wider.
The Auto Scaling band shows additional capacity waiting in the sidelines. This is added automatically when capacity thresholds are hit, well before users notice slowness.
This approach has an added benefit of providing inherent robustness in the face of failures. Looking at the above diagram, redundant application and database servers can live in physically separate locations. In AWS, these are called Availability Zones. If a server fails, the system automatically and transparently shifts load to healthy ones, while quietly bringing online and introducing healthy replacements. This works even if an entire datacenter location has a disaster. Users do not see an outage, and business operations continue uninterrupted.
AWS Virtual Private Clouds are made up of real data centers with all of the expected security capabilities of traditional corporate data centers. Physical networks can be designed with the same security best practices used internally.
AWS provides very fine-grained security controls for restricting users allowed to access an environment. Multi-Factor Authentication is standard, and a comprehensive audit trail of any activity is centrally recorded across the platform. Each component within the infrastructure exposes logs that can be watched by a wide range of security auditing tools for proactive security monitoring.
Highly Manageable Environments
One subtle aspect of deep virtualization from the servers all the way to network infrastructure is that environments can be essentially built with code. Infrastructure patterns for networks, failover regions, application and database servers etc. can be crafted as environment building blocks. Entire new environments, even in different parts of the world, can be described by assembling and configuring these modules into a form of environment specification. With one key difference from normal specification documents: these can be “printed” into real environments in minutes. This is the DevOps holy grail, allowing smaller teams to effectively produce and support larger environments. It removes human error from the process and allows the hard work of design to be done once and leveraged repeatedly.
Hopefully this post helps clarify some of the reasons to give the AWS cloud a very serious look, if you haven’t already. Not long ago, building distributed applications meant also building server farms. Picture trucks backing up to a building you own and unloading servers, cables, networking gear and air conditioners that needs to be setup, configured and maintained by a small army of techies.
Few clinical organizations truly want to build expertise in data center management. GxP compliant cloud infrastructure represents an enormous opportunity to improve application quality and experience for users, allowing us to focus on and invest in our strengths: all things clinical!