Selectel reviews Managed Kubernetes development in the first half of 2026

On 27 July 2026, Selectel published a review of Managed Kubernetes development during the first half of the year. The provider added dedicated-server worker nodes with 10 and 25 Gbit/s inter-node networking, launched a multi-zone cluster, expanded node configuration through user data, and introduced Envoy Gateway and audit logs.

These updates address enterprise infrastructure directly: microservice performance, resilience to a site failure, ingress traffic management and administrative accountability.

10 and 25 Gbit/s inter-node networking

Selectel now offers dedicated-server worker nodes with 10 and 25 Gbit/s networking between nodes. Higher internal throughput is particularly relevant to distributed databases, ML pipelines and services that exchange large volumes of data between pods.

It does not guarantee an automatic application speed-up. Before migration, a team should measure:

• the share of east-west traffic in the total workload;
• latency between nodes and storage;
• data-replication throughput;
• network-policy and ingress-layer limits;
• application behaviour when one node becomes saturated.

A multi-zone cluster

In the ru-6 region, a cluster type can distribute control-plane and worker nodes across separate data centres. This reduces dependence on one site, but it does not make an application disaster-resilient by itself.

Databases and queues need their own replication rules, applications need appropriate anti-affinity and topology spread constraints, and external integrations need redundant routes. A zone-loss scenario should be tested regularly against a disaster-recovery plan, including DNS, load balancers and dependent APIs.

Envoy Gateway and action auditing

Selectel replaced Ingress NGINX with Envoy Gateway in its Managed Kubernetes application catalogue. Before moving production traffic, teams should verify annotation compatibility, routes, TLS configuration, request limits and service-publication policies.

Audit logs record the creation, deletion and updating of clusters and node groups. These events should feed a central log store or SIEM, retain the initiating identity and change request, and follow an explicit retention period.

Node configuration through user data

Managed Kubernetes on dedicated servers now supports user data as cloud-config or a bash script. This can make worker configuration repeatable, but it requires the same controls as other infrastructure code: versioning, secret scanning, idempotence and testing on an isolated node group.

After configuration, teams need visibility into services, networking and application health rather than only script status. Our guide to infrastructure observability covers metrics, logs and traces.

Business conclusion

Selectel’s updates strengthen four aspects of a container platform: performance, resilience, traffic governance and auditability. Enterprise teams should evaluate them together with application architecture, SLA requirements and a tested recovery scenario, rather than as isolated cloud-console features.

Source: Selectel’s official corporate blog on Habr, published on 27 July 2026; the original source is linked in the article metadata.