Comic Relief: Multi-Cloud Microservices with Cloud Foundry for Massive Scale E-Commerce
Comic Relief is the poster child scenario for the value of Cloud computing – The organization experiences barely any demand for 364 days of the year, but an enormous volume on their flagship Red Nose fund raising day, a demand that is serviced by over 14,000 call centre agents and generates 800,000 donations, up to 400 a second at peak times.
And they must still achieve this via a relatively small budget. Meeting that demand via a traditional IT approach of buying and hosting the required capacity of their own servers that sit idle for 364 days would be financially impossible.
This MSP Mentor news about a major AWS outage defines the obvious business case for using multiple Cloud providers this way. Even the tier one providers experience outages and for revenue-critical systems this downtime means millions of lost dollars. Operating across two or more other Cloud providers eliminates this risk.
Early research and innovation focused on API-level integration across Clouds, such as this 2014 academic assessment of Multi-Cloud APIs, as a way of avoiding ‘Cloud lock-in‘. Telco industry body the TMF undertook research to define a use case for SDN/NFV, through their Multi-Cloud SDN project where vendors like Microsoft contributed considerable intellectual input, documented through this Sochi Olympics case study. Their Multi-Cloud Reference Architecture offers this as a repeatable, detailed blueprint guide.
Since then we’ve also had the Docker-led explosion of containerized deployment, contributing greatly towards this goal through enabling ubiquity and portability, with vendors also responding to the trend, such as VMware launching their Cross-Cloud Architecture.
This presentation from Nane Kratzke offers a detailed implementation blueprint:
Comic Relief utilizes the Cloud Foundry multi-cloud capability, and describing it as a polyglot approach, further designs in resilience of all components such as also using multiple connectivity and different payment providers, as well as operating across both AWS and Google for Cloud hosting.
The scenario employs a microservices and stateless architecture to enable the decoupling required and distributes shards across multiple Cloud instances, making possible unlimited horizontal scalability on demand.
The entire platform, infrastructure and applications, fully tested, can be deployed to the separate IaaS providers across three continents in 90 minutes, achieved by a team of only five. There is no need for any proprietary, high spec infrastructure or licences.
Microservices is ideal for this type of intensely highly volume, online ordering requirement for e-commerce. Chris Richardson, Founder of the original CloudFoundry.com, references an e-commerce scenario to define a best practice approach for decomposing legacy monolith applications into a microservices architecture.
Furthermore microservices are an ideal architecture for implementing and exploiting APIs, to better integrate e-commerce systems and build a Platform business model.
For example Api2cart focuses on the role they play in enabling integration of shipping service providers, like UPS or Canada Post, as well as specialist ventures like Shippo and Postmen, with ProgrammableWeb listing over 90.
Sensidia offers a vendor solution for microservice API enablement, describing how this can be used to integrate back-end legacy systems.
This combination of innovations offers a complete blueprint for e-commerce sites looking to modernize their platforms for massive scalability and agile DevOps development.