AllSaints: Soaring to heavenly site speeds and savings with Google Cloud
About AllSaints
From a single store off London’s Carnaby Street, it took just two decades for AllSaints to become a major global brand for menswear, womenswear, and accessories in more than 27 countries worldwide.
Tell us your challenge. We're here to help.
Contact us
About Pythian
Pythian is a global IT company, Google Cloud Premier Partner for data analytics, cloud migration, infrastructure, machine learning, and 2018 Google Cloud Data Analytics Partner of the Year.
Global fashion retailer AllSaints boosted online sales by moving to a microservices architecture on Google Cloud that cuts costs by half and improves page load times by 32%.
Google Cloud results
- Deploys new code four times faster, in less than five minutes instead of more than 20
- Accelerates page-load speeds by up to 32% on retail website, even during traffic peaks
- Provides a cloud platform that facilitates the creation of a retail platform
- Helps to create consistent development environments through Google Kubernetes Engine and containerization
- Smoother, faster site experience for customers, helping to boost conversion by more than 17%
Cuts infrastructure costs by 50%
Founded in 1994, AllSaints is a truly global brand, as popular in East Asia as it is in Europe or North America. Now in its 25th year, the fashion retailer shows no signs of slowing down, setting new internal records with its revenue growth. Online sales are one of the biggest contributors to that success, and with more than two million visits per month to the AllSaints website, delivering a smooth, fast online experience is a key challenge for the company.
“There’s no such thing as a perfect forecast in our industry,” says John Bovill, Executive Consultant Digital and Technology at AllSaints. “We make projections for rates of business in normal periods, but it’s very hard to predict how sales will rise during peak demand, especially online. Our need for infrastructure often doubles at those times, but we only actually need those servers for a very short period, such as Cyber Weekend.”
“Responsive websites and fast page-load speeds are critical for the mobile, connected customer. We keep things fast by using Google Cloud to provision capacity instantaneously, whenever it’s needed, to keep customers happy and sales up.”
—John Bovill, Executive Consultant Digital and Technology, AllSaintsAs an established luxury brand, AllSaints knows that its customers value consistency and a seamless online experience. Shortfalls in infrastructure capacity compromise that consistency by causing site instability and degraded performance. AllSaints met that challenge by providing more than 60 servers: more than enough for peak demand. Outside of peak periods, however, almost half of those were idle, making the solution expensive. More importantly, installing additional capacity was a slow process, delaying structural change such as expansion into new territories or the deployment of new features. AllSaints made a strategic decision to optimize the responsiveness and reliability of its retail operations by migrating to cloud infrastructure.
“Responsive websites and fast page-load speeds are critical for the mobile, connected customer,” says John. “We keep things fast by using Google Cloud to provision capacity instantaneously, whenever it’s needed, to keep customers happy and sales up.”
Halving the number of servers needed, with intelligent autoscaling
As online retail develops, companies need to respond to fast-moving and unpredictable changes in customer behavior and demand. That’s why AllSaints made moving to the cloud a strategic goal in 2014, as it looked to make its services as responsive and adaptable as possible to meet the challenges of a changing world.
“We started by deploying Google Workspace for our internal communications,” says John. “It was a deliberate strategy to get people used to cloud-based tools, both in stores and in offices. And at the same time, we started the switch to microservices.”
After experimenting with different systems for orchestrating containers, AllSaints opted for Kubernetes in 2016, using a mixed environment comprised of Google Cloud and another cloud provider. But the team soon realized that using Google Cloud alone would reduce costs while enhancing scalability and decided to “draw a line in the sand” to build new apps and services in Google Cloud alone. It began preparing to migrate all legacy from its split architecture over to Google Cloud, with support from Pythian, which provided key Google Cloud expertise. “We were moving 60 individual services, not just one application,” explains Andy Dean, Technical Operations Manager at AllSaints. “The interdependencies between them meant that it made more sense to move them all at once, and that took a lot of planning.” When the migration finally began, it was completed within a week.
The idea was to migrate “big bang,” or all in one go, and avoid any network latency between cross-service communications. The infrastructure team set up each technology type it needed, that is, Cloud SQL via MySQL on Compute Engine, hosts for Kubernetes, and Cloud Memorystore. Meanwhile, the dev team began re-coding to ensure that each service was compatible with its new Alpine images on Docker. Once the teams completed their setup of a fully functional dev environment in Google Cloud, they tested the build process to deploy into a test and production project, and then gave the sign-off.
“It was the biggest infrastructure change we’d made in the history of the company, so one of our goals was that nobody notice the change,” shares Andy. To achieve this, his team took the approach of working closely with Google Cloud and Pythian on code improvements and changes to applications. “We got things stable within the first week, which was crucial for us,” he explains. “In that week, we moved 60 individual services, including our ERP tills, to a microservices cloud environment.” After the move was complete, the project team spent an additional week making sure no stone was left unturned. “The support we received from Google Cloud throughout the migration helped make this happen. Before we started, I was nervous about the process. But I was happy to be proved wrong.”
“We monitor the architecture using Stackdriver, but Google Kubernetes Engine really looks after itself. The self-healing aspect of Google Kubernetes Engine means we no longer have to make time to restart some of the trickier VMs. Scaling is now automatic, and so are key maintenance tasks.”
—Andy Dean, Technical Operations Manager, AllSaintsNow AllSaints has halved the number of servers it runs from 60 to 30, relying on Compute Engine and Google Kubernetes Engine autoscaling to meet the needs of additional traffic. By eliminating the need for additional servers to meet spillover, the company cut infrastructure costs by 50%, while improving performance during periods of peak demand. Effective monitoring through Stackdriver helps the team proactively manage the architecture and reduce incident resolution times. And should incidents occur, AllSaints has improved disaster recovery by replacing its reliance on a single data center with backups on the globally distributed Google network.
“We monitor the architecture using Stackdriver, but Google Kubernetes Engine really looks after itself,” says Andy. “The self-healing aspect of Google Kubernetes Engine means we no longer have to make time to restart some of the trickier VMs. Scaling is now automatic, and so are key maintenance tasks.”
Faster development that responds to real needs
Unlike many of its competitors, AllSaints develops all of its customer-facing services in-house, from electronic point-of-sale systems in store, to the AllSaints app. For the company’s developers, moving to microservices on Google Kubernetes Engine was a chance to build a new CI/CD pipeline using Jenkins on Google Cloud and Terraform. By installing the new pipeline, the team has cut the time it takes to deploy a change by more than 75%, from more than 20 minutes to less than five. And the simplicity and speed of creating test environments in Google Cloud also means developers can test code on the same infrastructure that it will run on in production, minimizing the potential for problems.
“Before, we couldn’t confidently say a bug was fixed until we actually tested it in production. Now we can deploy code in test environments that exactly mimic production,” says Andy. “The improved CI/CD pipeline means we can update our services every day, with a shorter lifespan on bugs and minimal disruption. That makes us more responsive to customer needs, more proactive. And that's exactly what we’re trying to achieve.”
Consolidating efficiencies on a single platform
In its first month, the new platform produced spectacular results, as reported by the AllSaints team. Infrastructure costs are down by 50%. Page-load speeds are 32% faster, while conversion online, even by the strictest metrics, increased by 19% in the US and 17% in the UK, an increase John attributes directly to the fast load speeds.
“Strategically we are looking to maximize our usage of Google Cloud, driving this and associated technologies to provide the best possible AllSaints experience for our customers.”
—John Bovill, Executive Consultant Digital and Technology, AllSaintsNow the team is preparing to deploy Istio to connect and monitor microservices, while looking into ways to leverage data with BigQuery. And the migrations are set to continue into 2021, as backend and internal systems, from ERP to point-of-sale, are containerized and moved to Google Cloud.
“Strategically we are looking to maximize our usage of Google Cloud, driving this and associated technologies to provide the best possible AllSaints experience for our customers,” says John.
Tell us your challenge. We're here to help.
Contact us
About AllSaints
From a single store off London’s Carnaby Street, it took just two decades for AllSaints to become a major global brand for menswear, womenswear, and accessories in more than 27 countries worldwide.
About Pythian
Pythian is a global IT company, Google Cloud Premier Partner for data analytics, cloud migration, infrastructure, machine learning, and 2018 Google Cloud Data Analytics Partner of the Year.