DZ BANK unlocks 70% toil savings and 90% cost savings with a Cloud Run-first approach

Editor’s note: DZ BANK is the second largest bank by assets in Germany. In this post, Cloud Engineer Tim Harpe from DZ BANK shares how migrating to Google Cloud resulted in spectacular efficiency gains and cost savings.

DZ BANK chose Google Cloud to accelerate its digital transformation because Google Cloud offers cutting-edge technology, top-tier expertise, and meets the rigorous security and compliance standards demanded by the financial sector. In a short amount of time, we have containerized and migrated some of our most business-critical applications to Google Cloud and Cloud Run, achieving 70% toil reduction, 90% cost savings, and unlocking new capabilities through integrations with leading services like BigQuery.

As a result of this success, we're proud to be recognized as one of the winners of the Google Cloud Customer Award 2023 in the Financial Services Industry category!

Our modernization journey

As a 150 year old institution, DZ BANK has a lot of legacy infrastructure, much of it still running on-premises as virtual machines (VMs), with over 55,000 CPUs. On-prem VMs require a lot of effort to manage, and they're expensive. A few years ago, we decided to modernize our stack to reduce our overhead, increase developer velocity, and strengthen our competitiveness.

Around 90% of our application stack is developed in-house and managed by us. To modernize each app, we first containerize it and then deploy it on either Cloud Run or Google Kubernetes Engine (GKE).

Cloud Run is a great fit for many of our apps, especially ones that experience spikes in traffic. Since Cloud Run can quickly scale to thousands of containers, we don't need to provision for peak loads. We also have many smaller internal workloads that are not used during off hours. Now, we can scale them on request with Cloud Run, scale to zero when they're not being used, and only pay for the infrastructure when it's serving customer traffic. In the past, we separated each of these workloads into their own individual VMs that ran 24/7.

For apps where we want more control over the underlying hardware, we put them on GKE. The first app we migrated to Google Cloud was newly developed in-house to calculate customer risk and creditworthiness. It ran every night and scaled up to 60,000 CPUs. GKE enabled us to fine-tune scalability and hardware settings to exactly meet our needs.

Innovating for the future

Moving forward, DZ BANK is taking a Cloud Run-first approach with all newly-developed apps to help us innovate for the future. One initiative we have within the bank is to be more sustainable and a more responsible citizen of the world and we're now considering ESG (Environmental, Social and Governance) criteria in our credit processes. As part of this commitment, we developed a new app, running on Cloud Run, which evaluates creditworthiness based on ESG scores assigned to businesses.

Selecting Cloud Run has accelerated our development process. Cloud Run's built-in integrations for Cloud SQL, Secret Manager, and Cloud Load Balancing allowed us to create a simple architecture. Running on Google Cloud also makes it easy to access our custom data lake, which is powered by BigQuery.

From the perspective of our platform team, we also appreciate Cloud Run's approach to simplifying security. As the central bank for cooperative banks in Germany, DZ BANK must continuously meet stringent compliance and security standards. Cloud Run offers a strong foundation with encryption and built-in access control, plus integrates with the broader Google Cloud ecosystem of security products like customer-managed encryption keys (CMEK) and VPC Service Controls. This enables us to maintain control over our data while minimizing the risk of exfiltration.

Results and what's next

We are now three years into our modernization journey with Google Cloud, and we've seen significant benefits. Running containers on Cloud Run instead of on-prem VMs eliminates OS and infrastructure maintenance, delivering a 70% toil reduction. In addition, the pay-per-use pricing model has reduced our infrastructure costs by 90%.

We look forward to further partnering with Google Cloud in our journey to further modernize our fleet while also building out new innovative apps for the future.

^{This blog post received contributions from various people. In particular, we would like to thank Latif Ajouaoui for strategic insights, Yuriy Babenko for technical oversight, and Wietse Venema for reviews.}