APSIS One Architecture
Rikke Søndergaard avatar
Written by Rikke Søndergaard
Updated over a week ago

APSIS One's Architecture Overview

APSIS One is a combined marketing platform and CDP, designed from the ground up for making data-driven, omnichannel marketing on a massive scale simple and accessible to everyone.

It enables customers to create dynamic, personalized experiences for consumers. To do so, APSIS One ingests and stores billions of data points from both anonymous and identified individuals, connects the dots between the data points, and enable communication across multiple channels including email, SMS, web, mobile apps, social media and more.

All data and services in APSIS One are hosted on Amazon Web Services' industry-leading cloud infrastructure, running in strictly secured Virtual Private Cloud networks. We use an application-layer security model based on the OAuth2 framework to authenticate and authorize calls in every layer, including between our internal services. Every request is tracked with auditable IDs, and logs are redundantly replicated to secure tamper-proof accounts.

All services are load-balanced and set up redundantly across multiple data centres to ensure robustness.

Over 50 on-call engineers across 4 geographic locations monitor every service 24/7 using automated monitoring and dashboards. All data storage is based on an at least triple-redundant storage technology, and backed up at least daily. APSIS One has full disaster recovery plans in place and this is exercised at least on an annual basis.

Illustrative technologies we use include:

  • Native AWS technologies such as EC2, ECS, Fargate, RDS, Aurora, DynamoDB, AWS ElasticSearch, ElastiCache Redis, Athena, Lambda, S3.

  • Database technologies such as MySQL, PostgreSQL, and some NoSQL technologies.

  • Our Front-end stack is built on Angular 15.x with NgRx and of course involves HTML, JavaScript, CSS.

  • Our Back-end services are built using a mix of technologies including Golang, NodeJS Typescript, Java, and .NET core and typically either running as Lambdas or containerised with Docker.

All deployments are automated including batteries of automated unit, integration, and acceptance and smoke tests, and all infrastructure is provisioned using infrastructure-as-code through AWS CloudFormation.

Did this answer your question?