ifesca GmbH from Ilmenau, Thuringia, offers software solutions for energy-intensive industries and the energy sector. The solutions support decision-making processes in the area of forecasting and optimization. The AI-based, intelligent forecasting system supports energy supply and demand decisions by providing fully automated real-time forecasting.
The project at a glance
- Choice between AWS and Google Cloud Platform (GCP) based on scalability, cost efficiency, and ease of use
- Development of "lift & shift” migration and cloud-native refactoring scenarios
- Evaluation of cloud providers based on four proof-of-concept (PoC) implementations
- Enablement of the internal development team for AWS and GCP
Initial situation
ifesca 
operates its existing platform with a medium-sized provider for managed Kubernetes. The scaling possibilities and the pricing model are currently too sluggish compared to the highly dynamic load peaks.
In the future, ifesca expects millions of views per month, with a very elastic profile, i.e. steeply rising and falling traffic with pronounced peaks and valleys. These requirements do not fit the current setup, so ifesca wanted to evaluate the move to a public cloud provider.
Solution
The overall evaluation criteria for the selection of the future cloud provider are sorted by importance: Scalability, cost efficiency, and ease of use. The evaluation took place in three phases over six months.
Pre-selection of the two most promising cloud providers
Implementation of proofs of concept (PoCs)
Evaluation and decision
Pre-selection of the two most promising cloud providers
At the beginning of the project, there was a choice between the three market leaders Azure (Microsoft), AWS (Amazon) and GCP (Google). We presented ifesca with a comparison of the service offerings of the three providers and experiences from our projects at codecentric. On this basis, the decision was made in favor of proofs of concept in AWS and GCP.
ifesca chose AWS because the cloud provider offers many managed services and because its long-standing market presence promises a stable platform. The range of IoT services was another plus. Google Cloud Platform (GCP) was selected as the second candidate over Azure because it had the best Kubernetes offering at the time of the project. In addition, the platform is strongly positioned in the field of data analytics, which is an important topic for ifesca in the future.
Implementation of proofs of concept (PoCs)
codecentric provided the cloud resources for ifesca. Following onboarding for AWS or GCP and with continuous support from codecentric IT consultants ifesca's internal developers created two PoCs for each cloud provider : “lift & shift" and "Cloud-native refactoring".
The “lift & shift” PoC was used to test how quickly the current Kubernetes setup can be moved to the cloud. The PoC was also to provide an initial indication of the cost savings and be completed within just a few weeks.
ifesca plans to move the software to cloud-native serverless technologies in the medium term because using Kubernetes incurs high costs for unused computing resources and a high level of unproductive maintenance for the customer. They expect this to result in significantly lower costs per call and less maintenance. This was tested in the "cloud-native refactoring” PoC.
In total, four PoCs were realized:
PoC 1 : Lift & shift of the existing Kubernetes setup to AWS EKS (Elastic Kubernetes Service), migration of the database to AWS Aurora
PoC 2: Cloud-native refactoring of the application on AWS Lambda and Step Functions after AWS Fargate did not work
PoC 3 : Lift & shift of the existing Kubernetes setup to GCP GKE (Google Kubernetes Engine), migration of the database to Cloud SQL
PoC 4 : Cloud-native refactoring of the application on GCP Cloud Run and Cloud Workflows
Evaluation and decision
After our experience with the PoCs, we evaluated the cloud providers based on the criteria of scalability, cost efficiency and user-friendliness.
Both cloud providers were on a par with the criteria of scalability (i.e. processing simulated requests in a reasonable time). GCP was ahead in terms of user-friendliness / developer experience, ifesca's developers liked the clarity of the Web UI, the consistency of the services, and the quality of the documentation.
It is true that GCP had a cost advantage for databases and computer capacities as well as Kubernetes. However, the serverless stack at AWS showed a measurable, significant cost advantage, making AWS considerably cheaper after only a few million calls per month. AWS thus proved cost-effective, which is ultimately the more important criterion in terms of user-friendliness. ifesca therefore chose AWS as its future cloud provider.
Result
The existing software installation was not designed for ifesca's growth plans. We therefore evaluated a migration to AWS or GCP based on customer requirements. Together with the customer, we create four PoCs on AWS and GCP in "lift& shift" and "cloud-native refactoring” scenarios. GCP has the lower fixed costs for databases and compute resources, but AWS is already catching up with the lower-priced serverless stack after relatively little traffic. Since AWS and GCP are on a par in terms of scalability, and ease of use is secondary to cost efficiency, lower costs were the deciding factor for ifesca's decision to choose AWS as its future cloud provider.
By enabling the development team through our employees as part of the project, ifesca can start the productive migration to the cloud at a time of its own choosing.
Clemens Rosenbauer
Chief Technology Office, ifesca GmbH
Choosing a cloud provider was an important decision for our company moving forward. Thanks to codecentric's expertise, we were able to quickly achieve a result that best suits our requirements – knowledge transfer and empowerment of our employees included.
Clemens Rosenbauer
Chief Technology Office, ifesca GmbH
Do you have more questions about the project?
Would you like to learn more about the project? Are you interested in a similar solution for your company?
Service Lead GenAI
Other codecentric AG projects
Find out about other successful projects that we have completed with our clients. Perhaps you will find ideas here for a use case in your own organization.
Niklas Haas
Service Lead GenAI