-min.avif)
Introduction
Bayerische Motoren Werke AG (BMW) is one of the largest premium car manufacturers globally. Offering speed, agility, and a premium experience with the help of innovation, which is at the very core of BMW’s identity. However, providing a superb user experience on the web with fast loading times across a multitude of countries is a huge challenge. This involves taking into account numerous stakeholders as well as country-specific technical challenges.
"Speed Kit has been rolled out for 118 countries and achieved a 1.5x faster LCP overall."
Challenge
Solution
Speed Kit uses the new service worker technology to apply innovative caching algorithms for faster delivery of both static assets and highly dynamic HTML. Speed Kit therefore massively reduces latency through consistent browser caching and global edge caches. As a result, Speed Kit solves the Time to First Byte (TTFB) bottleneck leading to faster rendering and a much improved user experience. In addition, Speed Kit also accelerates third-party services, optimizes the loading order of critical resources, and applies automatic image optimization.
Speed Kit is an official ACE eligible technology partner and independent software vendor (advanced tier) of the AWS Competency Partner Program "Accelerate" as well as the Workload Migration Program. Our technology is powered by 13 different AWS services:
- Elastic Kubernetes Service (EKS): We use EKS to schedule and orchestrate our Speed Kit applications on EC2 instances.
- Simple Storage Service (S3): Speed Kit stores cached assets in S3 buckets. Additionally, we store RUM (Real User Monitoring) and PI (Performance Insights) data in S3 buckets for analytics purposes.
- Elastic Compute Cloud (EC2): Our workloads are hosted on EC2 instances, as managed through EKS.
- Kinesis Data Streams: Kinesis Data Streams are used to ingest RUM and PI data from Speed Kit, which is then consumed by Amazon Managed Service for Apache Flink. Fastly also streams access logs to our Kinesis Data Stream, which are similarly consumed by Amazon Managed Service for Apache Flink.
- Athena: We use Amazon Athena to query data stored in AWS S3 buckets, enabling performance insights and asset preloading.
- Elastic Container Registry (ECR): Docker images are stored in ECR and deployed in our Kubernetes cluster, which is managed by EKS. For third-party application images, we use the pull-through-cache feature of ECR.
- Elastic Container Service (ECS): We build Docker images using a service hosted on Amazon ECS.
- Route 53: Route53 is used to manage DNS records.
- Simple Email Service (SES): We use SES is used for sending transactional emails.
- Managed Streaming for Apache Kafka (MSK): We use MSK to facilitate consuming and producing custom event records for different applications.
- DynamoDB: Speed Kit uses DynamoDB to store information related to our predictive preload feature.
- Lambda: We use Lambda to pre-render client side renderd pages to make that actual server side rendered.
- ElastiCache (Redis OSS): We use ElastiCache to store our Bloom filter, which checks whether an asset is present in our cache.
-min.webp)
Click the image to open the architectual diagram of Speed Kit with AWS
Results
Since Speed Kit is implemented via a JavaScript tag in the HTML template, the speed uplift can be A/B-tested. Through the use of real-user monitoring, we obtained statistically significant improvements across all web performance metrics.
Largest Contentful Paint (LCP):
The consistent uplift in the Google Core Web Vitals like the Largest Contentful Paint (LCP) across all contintents confirmed a substantially improved user experience. Overall, Speed Kit saves 64% of unnecessary image data for over 196 million monthly images delivered to users.
