Overview.

AME Kubernetes - managed platform

Avisi Managed Environment Kubernetes is a managed Kubernetes platform across multiple cloud providers. It is the core of our Avisi Managed Environment platform. AME Kubernetes clusters are fully compatible with upstream Kubernetes.

Service Levels

ItemDescriptionBasicPremiumEnterprise
SLA windowBusiness hours24x724x7
Scheduled maintenanceCustom maintenance windows (week days only)
Workload availabilityAvailability requested workloads. Not applicable for BYOI.99,9%99,9%99,99%
ClusterControl plane availability Single Control plane instances99,9%99,9%99,99%
Multi-zone clusterDeploy a cluster across multiple availability zones-99,9%99,99%

Features

FeatureDescription
Security hardeningClusters are hardened based on guidelines from the Cybersecurity and Infrastructure Security Agency (CISA).
Security monitoringAutomated security monitoring against your clusters
Audit trailingAudit logging is available by default in all clusters, regardless of SLA
MonitoringIntegrated support for Prometheus, part of our Avisi Application Observability
LoggingUse Loki in any cluster with long term retention. Part of Avisi Application Observability
ScalingEasily scale your cluster up/down. Run only what you need.
ComplianceAvisi Cloud has a SOC2 Type II report and is ISO27001-certified.
Identity managementSingle Sign-on using personal accounts and full RBAC capabilities. Support for Multi-Factor Authentication (MFA).
Upstream & Conformant Kubernetes

AME deploys an upstream Kubernetes and remains close to the behaviour of standard Kubernetes. All clusters pass the Kubernetes conformance tests by the CNCF.

Every new release by AME must pass these tests before they are published, along side various additional end-to-end tests by Avisi CLoud itself.

Automated cluster nodes

Nodes are managed in a fully automated way. We use immutible infrastructure princples to safely and reliably patch systems.

Should a node become unreachable due to hardware failure, or any other number of reasons, it will be automatically replaced by a healthy node to ensure available capacitity within your environment.

Load balancers

A way to dynamically route traffic from the public internet to your application. Implementations may differ between cloud providers.

Persistent storage

Easily persist data within your cluster using external storage. We support native implementations in each public cloud we support and support Ceph for private cloud implementations.

VPC networking

Each cluster is fully isolated from any other cluster and/or customer. No two clusters can reach each other over the network.

Cluster networking

We support multiple plugins for in-cluster networking. By default we deploy Calico.

Support is available for weave net and Cillum. With all implementations, we support network policies.

Feature compatibility across providers

Since AME supports multiple providers, certain implementations may be different across clouds.

Avisi Cloud will only support providers that has the following basic functionalitity supported;

  • Provisioning of Load balancers - a way to dynamically route traffic from the public internet to your application.
  • Provisioning of Persistent Volumes - easily persist data within your cluster using external storage.
  • VPC Networking - fully isolated networking from any other cluster.

How to

Limitations

  • The amount of clusters you can create is limited per organisation. The same counts for the amount of nodes within a cluster.
  • The Control plane is fully managed by Avisi Cloud. You cannot modify any settings other than available through the API.
  • When upgrading a cluster you can only upgrade to the next minor or patch version.

Cluster limits

  • Clusters can have up to 250 nodes per zone. Note that each organisation starts with a lower limit (5).
  • Pod limit per node is 110.
  • We only deploy containerd as our primary container runtime (e.g. no Docker).
  • It is not possible to disable the metrics-server add-on at this time.
Available memory per machine size

While you can provision machines with 8GB of RAM, you cannot use the entirety of this for your application. A certain part of this memory is used for the operating system and other system components such as the container runtime.

Machine sizeOverheadAvailable
4Gi300Mi3.7Gi
8Gi800Mi7.2Gi
16Gi1200Mi14.4Gi
32Gi1600Mi29.6Gi
64Gi2400Mi61.6Gi