“By being able to dedicate the new infrastructure to the critical delivery of public facing weather services, we’re able to respond more quickly to changing events, deploy more rapidly and quickly scale with demand.Chris Beighton, Met Office Solutions Architect
Overall the engagement has been, and continues to be, a positive and collaborative one which is already enabling us to use our new Met Office apps to reach and engage with more users, helping them to make decisions based on world class, fast and accurate weather advice.”
The new cloud infrastructure was needed to support Met Office digital services. Demand for weather information can peak sharply during weather events such as storms or heavy snowfalls across the UK, so the infrastructure also had to be flexible, and respond to demand dynamically. Using traditional hosted solutions, it’s not always possible to deploy new resources for very short periods of time, without significant costs and delays. This led to the need for a trusted partner who could quickly deliver a robust and flexible infrastructure.
- Everything in AWS is an Application Programming Interface (API) which can be called programmatically. We built a series of templates which are used to quickly create multiple identical environments for product testing, development and the live production environment. Each of these environments is isolated from the others and has capability to deliver future applications that would be both Highly Available and Elastic in nature, scalable, and swift to deploy.
- Each environment was designed to be highly available, and is built across 3 Availability Zones (AZ) within an AWS Region, so that in the event of one, or even two AZ failures, Met Office’s services will still be available.
- Packer was used to speed up the responsiveness of scaling events to pre-build Amazon Machine Images (AMI) for the instances. This let us build in software updates, monitoring tools and auditing configuration into the images, reducing the start-up time for a new instance to a few minutes with code-deployment.
- Additionally, we configured this as a Jenkins job, so that changes to the AMI are kept in source control, and automatically built into a new AMI when changes are pushed to our code repository.
- In the event of any problems during the deployment, AWS CodeDeploy will ‘roll back’ the deployment and allow for log analysis to determine the problem. As all Elastic Compute Cloud (EC2) instances within the environment are in AWS AutoScaling groups, they are able to recover from failure within a few minutes and ensure that defined minimum levels of EC2 instances are always available to serve end users.
- We created a ‘shared’ environment for tooling common to all environments, such as Jenkins. We ensured that, in the event of a failure, a fully working Jenkins server is recreated and usable by application developers within 15 mins.
- We installed the trusted cloud security service of AlertLogic across the environment to ensure security is continually monitored and secured. This leverages the service orientated approach to security, combined with leading security tooling from AlertLogic, and the secure synergy between AlertLogic, Cloudreach and AWS.
As the Met Office has a remit to provide vital weather services to the public, it was important for all parties that we not only deliver a functioning infrastructure solution, but also one that closely fits their requirements for security, high availability and auditing. We liaised closely with their teams in Security, Application Deployment, Load Testing and Content Delivery to ensure that they not only understood what we were delivering, but also that we were able to deliver a flexible infrastructure that fulfilled all their different needs.
We delivered a robust Cloud infrastructure which now supports The Met Office’s services. The new infrastructure can respond to demand changes quickly, is highly resilient in case of any failures as well as meeting security requirements. Using the Cloud means that the Met Office do not have to maintain hardware that is unused for large parts of the year, but can instead add and remove resources quickly and easily as demand fluctuates, whilst paying as they use. The scripted nature of the environments means that testing new features can be done faster, without disruption to existing work.
About The Met Office
For over 160 years, the Met Office has led the field in weather and climate forecasting, fulfilling its role as the UK’s national weather service in order to protect the public, ensuring our wellbeing and prosperity. In order to better support their growing user base, they identified the need for a backend system through which they could receive and transmit large volumes of data. Cloud services were selected as the ideal infrastructure to support agile deployment and scalability requirements. It was recommended that The Met Office should speak to Cloudreach, as a specialist who could help implement that infrastructure within Amazon Web Services (AWS).
Related Case Studies
Cloudreach helped Aware build a machine learning pipeline with Amazon SageMaker to enable their internal data scientists to focus on innovation.
Cloudreach helped Dufry reduce their AWS bill by 66% with cost visibility and optimization.