Cloud.ru updates Kubernetes, PostgreSQL, DWH and VMware monitoring

On 9 July 2026, Cloud.ru published its digest of June platform updates. The infrastructure changes include new Managed Kubernetes components, manual PostgreSQL cluster suspension, gRPC load balancing, table-level DWH backups, DNS resources for Terraform and a VMware infrastructure monitoring API.

For an enterprise team, the relevant point is not the list of versions. The changes address routine operational tasks: faster container deployment, temporary cost reduction, wider infrastructure automation and additional data for unified monitoring.

Managed Kubernetes changes

Cloud.ru updated Kata Containers, Istio, Trivy Operator, K8s Cleaner, KEDA and External Secrets Operator plugins. Spegel has also been added to the service. It distributes container images between cluster nodes and can reduce dependence on an external registry when an image is already available inside the cluster.

Ingress Nginx can now enable PROXY protocol during installation and editing. This is useful when an application needs the original client address through a compatible load balancer, but both ends must be configured consistently. A mismatch can make incoming traffic unavailable instead of improving observability.

A newer plugin version should not be enabled in production by default. Teams need to check compatibility with the Kubernetes version, network model, security policies and existing manifests.

PostgreSQL, load balancing and the data warehouse

Evolution Managed PostgreSQL clusters can now be stopped and started manually. A cluster can remain stopped for up to 30 days, reducing compute charges while it is paused. This suits development, training and temporary environments, but does not replace deleting unused data or defining an environment lifecycle.

Advanced Elastic Load Balance adds gRPC support for Dedicated load balancers’ Backend Server Groups. Microservice teams may no longer need a separate route for this traffic, but they should verify HTTP/2, timeouts, health checks, maximum message size and long-lived stream behaviour before migration.

Advanced Data Warehouse Service now supports backups of selected tables, diagnostic-result export and manual statistics collection for hot and cold table data. Selective backup is useful for critical data sets, but a recovery plan still needs to cover dependencies between tables, schemas, permissions and external ingestion jobs.

Terraform and VMware monitoring

The Terraform provider has added resources for DNS endpoints, endpoint assignments and DNS Resolver rules. This allows hybrid DNS configuration to be stored as code, reviewed through version control and repeated across environments.

The VMware cloud platform now has an API reference for extended monitoring. The API exposes virtual-machine lists, resource identifiers, available metrics and metric values. This enables infrastructure telemetry to feed a common observability system without manual export from the provider console.

What to test

A pilot should keep the changes separate:

1. test image distribution and an unavailable external registry in a non-production Kubernetes cluster;
2. measure PostgreSQL stop and start times and verify backup availability;
3. run unary and streaming gRPC calls through the load balancer;
4. restore selected DWH tables into an isolated environment and validate consistency;
5. create DNS resources from a clean Terraform state and import existing ones;
6. compare VMware API metrics with the provider-console values.

Our IT infrastructure observability guide provides a monitoring and alerting model. The guide to resilient API integration architecture is also relevant to DNS automation and external dependencies.

Practical conclusion

Cloud.ru’s June package consists of small but applicable infrastructure improvements. Spegel for container images, managed PostgreSQL suspension, gRPC balancing, selective DWH backups and the VMware monitoring API are the most relevant changes.

The digest does not provide independent performance measurements. Each feature therefore needs testing against the customer’s workload, while plugin upgrades should pass through a test cluster and a controlled change window.

Source: the official Cloud.ru digest dated 9 July 2026.