Less microservices, more control -

PHAINA's system landscape follows the single-tenant approach, meaning that PHAINA's customers use a shared infrastructure but have separate environments. Before switching to Flying Circus, PHAINA used the microservice landscape and Database for PostgreSQL from Microsoft Azure with a Terraform deployment and had designed its services as an infrastructure as code solution. However, this soon presented the challenge of an increasing number of individual services, which led to considerable complexity in the system landscape. A microservice architecture requires the orchestration, monitoring and scaling of each individual service, which continuously increased the administrative workload for the PHAINA team.

At the same time, the dependencies between the services had to be managed precisely in order to guarantee fault tolerances and ensure consistent data processing of the search queries. By processing queries across multiple microservices, a single failure or increased latency in one service could impact the overall user experience. Troubleshooting and problem diagnosis were made more difficult by the large number of interlinked services, as the cause of a problem could often be spread across multiple services. An unrecoverable error case with a lack of technical support from the old provider, which affected the end customer, prompted PHAINA to take a fresh look at the infrastructure used and the internal workflows. All in all, the challenges led Niclas Köhler, CTO of PHAINA, to rethink the current strategy and look for new solutions to make PHAINA's solutions more stable and effective.

PHAINA came to the Flying Circus as part of a proof of concept (PoC) to test our application operation services and the quality of our technical infrastructure. Our PoC offer enables new partners to fully experience and test our services on a real project. As part of the PoC, PHAINA transferred its existing Terraform deployment to us. Our experts went through the deployment step by step to determine the requirements and the new deployment design and to create an executable system landscape for the PHAINA applications.PHAINA's system landscape consists of various components, with application containers for the APIs, frontends and the backend based on Docker playing a central role in the single-tenant structure. For the databases, we have combined a PostgreSQL database with Redis. PostgreSQL serves as a primary, persistent database that reliably stores structured data. Redis acts as a fast in-memory cache and message broker that stores frequently used or temporary data in order to reduce latencies and relieve the PostgreSQL database. The system landscape is completed with an S3-compatible object storage, an NGINX for SSL termination, HAProxy for load balancing and DNS services from Flying Circus to simplify certificate handling and reduce complexity, also through fewer external dependencies. Finally, an external e-mail service is integrated into the deployment of the system landscape.

Deployment is now realised by Flying Circus via the Batou tool on its own cloud platform and offers a production and, if required, staging environment for every PHAINA customer. The application containers are created by the PHAINA team via a GitHub pipeline and received by the Flying Circus using Batou. In this process, we have already taken into account that different containers can be provided for both staging and the production environment, for example to test new functions or enable better debugging. During deployment, these are then automatically provided with important configurations, customer-specific features and environment variables, both with mounts (e.g. for image and media data) and with string variables. Niclas Köhler and his team are able to independently and easily provision new customers and infrastructure resources on the Flying Circus Cloud platform as required via the deployment.

Phaina opted for the Managed service level at Flying Circus, which offers experienced development teams like PHAINA a high degree of self-management with an expert (living ;-) co-pilot. We support PHAINA in the realisation of new ideas and ensure that these are transferred to Batou Deployment, which PHAINA can use in self-management for its customers. Our application operation at Flying Circus offers various SLA levels, providing a flexible choice for every project requirement. In managed projects, the response time can be reduced to up to 15 minutes on request. Our flat-rate SLA levels offer planning security for the budget and provide a clear and transparent basis for co-operation. Both sides benefit from cost control, financial predictability, quality and efficiency through this approach.

By switching to Flying Circus, PHAINA was not only able to significantly reduce its operating costs and benefit from a more personalised service, but also improve the return on investment (ROI) of its solutions. As a partner, the Flying Circus offers a German cloud platform that is ISO 27001 certified and therefore offers a high level of IT security and stability. PHAINA benefits from a data protection-compliant Infrastructure as Code solution, which enables fast release cycles, a high degree of self-management and continuous further development of its software solutions.

"I could go to infrastructure service providers that scale to the masses and therefore offer more favourable offers for CPU, RAM and hard disk space than the Flying Circus. But that wasn't the main reason why I decided to work with the Flying Circus. The real advantage is that your support and assumption of operational responsibility gives me so much freedom that I would actually have to offset it with my own manpower in our company in another hosting construct. If you look at it from a commercial perspective, the service price at Flying Circus definitely pays off," says Niclas Köhler, CTO of PHAINA.

Prospects:

We at Flying Circus are already looking forward to the next project with PHAINA. As part of our efforts to provide our customers with a secure and trustworthy environment for their applications, we are currently working on implementing privacy-compliant features for Large Language Models (LLMs) on our Flying Circus platform. Our goal is to ensure that our customers can realise the benefits of LLMs without compromising on privacy and security.