Our customer usecases

In 2023, Mistral AI entrusted us with the design and management of their multi-cloud Kubernetes (K8s) platform dedicated to the inference of their LLM AI solutions (“Le Chat,” API).

We optimized the containerization of applications and their packaging using Helm. Deployment on the clusters is carried out via GitOps with Flux.

The K8s clusters are deployed on Azure AKS and Google GKE using Terraform and Terragrunt. To ensure high performance, they incorporate AI-specific optimizations:

• for storage and deployment of large models
• on drivers to support the latest Nvidia GPUs
• on GPU-based node autoscaling, and more

The platform also includes databases capable of handling high load peaks. It is operated using modern monitoring technologies and dashboards tailored to AI-specific needs.

Since late 2023, we have been providing 24/7 operations and maintenance (O&M) for the entire inference platform.

Environment : Microsoft Azure, Google Cloud Platform

Technologies : Kubernetes (Azure AKS, Google GKE), Docker, Helm, Flux, Databases

In 2022, Oodrive faced technical and organizational challenges in managing its platforms and approached Enix to unlock this complex situation.

We first revamped the on-premise platform: re-architecture, updating and optimizing Proxmox infrastructures, upgrading HW performance of PostgreSQL databases, redesigning the Arista network backbone.

Then we supported them in their cloud-native transition:

• Docker and K8s training, apps containerization, Helm charts, Harbor image registries
• 100% automated GitOps chain with Flux and GitlabCI for the deployment of business apps and infrastructures
• Observability and metrology platform measuring QoS on Oodrive’s client services
• Custom K8s clusters meeting sovereignty and advanced security requirements (SecNumCloud audit)

Since early 2023, we manage the 24/7 operations of the Promox, Kubernetes, and Harbor clusters, the GitOps chain, and the business observability platform.

Environment : Bare Metal (on-premise), Proxmox, Arista

Technologies : Kubernetes, Docker, Helm, Harbor, Prometheus, Thanos, Ansible, Gitlab, Flux, PostgreSQL

Quarkslab is one of the leading French players specialized in IT security solutions. They approached Enix for our infrastructure and cloud-native expertise for projects with high security requirements.

To host a security-oriented SaaS service, we deployed a platform on OVHcloud HDS bare metal servers. It consists of our Proxmox Private Cloud service and tailored secure K8s clusters with a Talos Linux base (a minimalist and immutable OS designed for K8s). It also includes an OVHcloud Cloud Disk Array service (Ceph, block mode) and ZFS storage local to the Proxmox hypervisors.

We manage the 24/7 operations of the Proxmox and Kubernetes clusters with Cloud Native monitoring based especially on Prometheus, Thanos, and Grafana.

In parallel, we also designed and deployed an ultra-secure Kubernetes cluster in an air-gapped environment.

Environment : Bare Metal HDS (OVHcloud & air gapped), Proxmox Private Cloud (Enix)

Technologies : Kubernetes, Talos, Helm, Docker, Proxmox VE & PBS, Prometheus, Thanos, OVHcloud Cloud Disk Array (Ceph, mode block), ZFS, Min.io, PostgreSQL

Since 2020, the Pasteur Institute has been working with Enix for the deployment and daily operations of four K8s clusters managed in Gitops with Cloud Native monitoring (Prometheus, Thanos).

These clusters are used in self-service mode by scientists for their applications, intensive computing (GPU/Nvidia DGX), and their Data Lake Owey. They comprise over 60 nodes and rely on a local Harbor registry. To manage security in a multi-tenancy context, we have implemented network-level filtering (NetPol) and K8s resource validation policies with Kyverno. User authentication on the clusters uses the Institute’s centralized database. For scientists working on Machine Learning, we have also set up a Kubeflow platform.

We have also trained their teams in Docker and K8s and support them in containerizing their applications.

In addition, Enix has deployed and is managing a Ceph storage (block mode) and a log management platform based on ElasticSearch, Kibana, and Vector.

Environment : Bare Metal (on-premise), VMware

Technologies : Kubernetes, Kubeflow, Kyverno, Helm, GitlabCI, Harbor, Prometheus, Thanos, Ceph, ElasticSearch, Vector

Since 2020, Enix has taken over the entire hosting of Optimi’s platforms (a significant B2B player in CDN, web performance, and Cyberdefense) and has been operating them 24/7. Optimi can thus fully concentrate on its business without worrying about infrastructure and hosting issues.

Initially, we built them a secure dedicated private cloud by deploying our Proxmox Private Cloud solution on dedicated Enix servers (in Interxion & Telehouse datacenters). We are also in charge of the O&M of a Ceph Object S3 storage and a logging/search platform based on ElasticSearch.

Subsequently, we supported them end-to-end in their DevOps and Kubernetes transformation: training, containerizing apps & helm charts, Build and Run of Kubernetes clusters.

We have also set up and we are now managing daily their software factory so they can focus on their business development.

The hosted applications are critical for Optimi’s business and its clients, with strong SLAs 24/7 both on performance and service availability.

Environment : Hosting Enix, Proxmox Private Cloud (Enix)

Technologies : Kubernetes, Helm, Prometheus, Thanos, Proxmox VE & PBS, GitlabCI, Ceph (S3), ElasticSearch

Recognizing the major advantages of containerizing applications, the technical management of the WEKA group approached Enix for support in their migration to Kubernetes and Cloud Native technologies.

We initially assisted them in containerizing their applications. Then we built a 100% automated software factory based on GitlabCI that allows the various entities of the group to deploy applications quickly and autonomously. These applications are hosted on tailored K8s clusters (Rancher RKE base). We have been managing these platforms 24/7 since 2018.

For cost optimization and to simplify overall operational management, we recently migrated them from their historical hosting provider with VMware to our dedicated and managed “Proxmox Private Cloud” solution.

In parallel, we have set up and manage 24/7 the O&M of their MariaDB, MongoDB, and PostgreSQL databases as well as their logging platform based on ElasticSearch and Graylog.

Environment : Hosting Enix, Proxmox Private Cloud (Enix)

Technologies : Kubernetes, Helm, Gitlab, Proxmox VE & PBS, ElasticSearch, Graylog, MariaDB, MongoDB, PostgreSQL

Emencia is a web agency specialized and recognized in Python and Django for over 20 years. Additionally, it offers hosting solutions to some of its clients.

Emencia called on Enix to help redefine its hosting strategy and rationalize its infrastructures. We optimized costs and simplified operational management of the platforms by migrating them to our Proxmox Private Cloud offering and managed the transition without impacting its business or that of its clients.

Subsequently, Enix deployed the platforms under Kubernetes (Rancher RKE base). We have been managing them 24/7 since early 2022.

Environment : Hosting Enix, Proxmox Private Cloud (Enix)

Technologies : Kubernetes, Rancher, Docker, Helm, Prometheus, Thanos, Proxmox VE & PBS, Stockage NFS

The IT department of MyDataModels called on Enix to design and deploy a new 100% automated continuous integration and deployment chain adapted to its organization and the specificities of its software development.

We have notably implemented advanced features in their CI/CD chain based on GitlabCI, such as Feature Branches and security-related features.

We have also provided custom Docker and Kubernetes training to the development teams and those in charge of the Kubernetes clusters on Amazon EKS.

Environment : Amazon EKS

Technologies : Kubernetes, Docker, Helm, GitlabCI, ChartMuseum, Ansible