AWS Kubernetes with Ondat

The Kube Native Data Platform for AWS

Get Started

AWS Kubernetes persistent storage

AWS provides an ever-expanding range of tools and services for developers and has become a popular platform to run Kubernetes on. Customers of AWS can choose to build out their own clusters on EC2 instances or use EKS the managed container services that AWS provides. The other architectural decision that customers need to make is around stateful workloads like databases. Are they going to use RDS, the managed database service? Do they run their own databases? If they are running their own database, the question of what storage service they should use becomes key. Getting the right architecture is vital; the right architecture will allow your application to scale, perform as required, maintain uptime and do all of this for the right cost. Ondat is here to help.

Kubernetes with Ondat evolves how stateful workloads are executed, transparently solving common issues like data loss, data unavailability, or manual provisioning of volumes. Ondat implements a fully Kube-native integrated data service platform that allows you to productionize stateful workloads. Applications like MongoDB, Kafka, MySQL/MariaDB, PostgreSQL, InfluxDB, or ElasticSearch, among many other use cases, can run transparently with high availability, self-healing, data protection, encryption, or topology-aware data placement by running on Kubernetes with Ondat.
Ondat minimizes human interaction with data services, making the operations easier than ever before. Ondat and Kubernetes manage the workflow of data for you, so your operation teams no longer have to be application experts to run them safely. Ondat also minimizes operational costs by facilitating the deployment of workloads across different platforms.
You can run workloads in AWS using the same procedure as any other Kubernetes distribution – whether they are running on Google Cloud with GKE, AKS, or VMware in bare metal, or even OpenShift. Ondat abstracts the particularities and complexities of the different platforms, offering one unique user experience.

AWS EBS with Ondat

AWS Kubernetes clusters use EBS as the default storage backend. However, EBS being VM-centric has operational, financial, performance, and scalability challenges when it comes to container-centric storage. You can drastically improve the storage capabilities by using Ondat between Kubernetes and EBS.

Ondat enables EBS to perform optimally at what it is designed for – EC2 backend storage. Ondat is designed to manage container storage, and running EBS together with Ondat offers the best outcome for stateful workloads on Kubernetes running on AWS.

Reduce compute costs of stateful apps by up to 90%

As organizations continue to drive deployment, velocity, and new applications, stateful workloads hinder the acceleration, leaving organizations to consume managed databases at ever-increasing costs.

Ondat enables organizations to realize all the benefits of CI/CD and Gitops investment for their stateful applications, delivering a Kube-native experience for all containerized applications, while dramatically reducing infrastructure and operations costs.

Learn why organizations like Accenture, Amarena, Interconnect, and more trust Ondat to manage data for mission-critical Kube-native applications.

Ondat EKS Better Together - JPG Overview

Ondat and EKS: Better Together

Amazon Elastic Kubernetes Service (EKS) is a managed container service used to run and scale Kubernetes applications in the cloud or on-premise with the EKS Anywhere offering.

Many AWS customers rely on EKS to reduce the complexity of managing and running Kubernetes workloads. When running stateful workloads in EKS, Ondat provides a unique set of features that enhances the capabilities of the AWS storage services. Ondat provides additional features, performance, and cost benefits whilst using the underlying AWS storage service to work in a true Kube-native fashion.

Download Datasheet

The Kubernetes Native Data Platform for AWS

Ondat is a leading Kube-native data services platform. Layered on top of EBS, Ondat’s innovative data mesh delivers the missing resilience, performance and cost-efficiency needed to run stateful business applications at scale, as well as dramatically reducing costs for both storage and other compute resources.

Deliver a serverless experience in your Kubernetes environments with Ondat. Providing deterministic performance, data availability, and business continuity, whilst significantly reducing your EBS costs up to 75%.

Download Datasheet

Operational advantages of using AWS EBS with Ondat

EBS has a limit of 25-39 EBS disks per node, depending on the EC2 instance type, and it might take minutes to provision and attach. Ondat on the other hand, can provision, attach, detach and delete hundreds of volumes per node per second.
Ondat uses synchronous replication to keep ready-to-use copies of the data. So in case of failure, Ondat automatically switches the data paths transparently to the application keeping the data always accessible.
EBS volumes have to follow the Pod across the cluster – a behavior it is not designed for. As a consequence, relocation of workloads with EBS can have significant downtime, even requiring manual intervention sometimes. Ondat provides data ubiquity and Pod scheduling influence, so when Kubernetes reschedules workloads, Ondat provides the data access wherever it is required (this most often being where the data resides).
EBS volumes are restricted to one AZ, whereas Ondat implements topology-aware placement, offering multi-AZ data safety. Upon request per volume, Ondat can replicate synchronously across AZs. This improves the HA capabilities and keeps the data safer in case of failure.

Financial advantages of using AWS EBS with Ondat

Ondat virtualizes backend EBS volumes using thin-provisioned frontend disks

It handles the data from many applications and optimizes writes. The data is then stored in the EBS of the EC2 instance, or any specific disk provisioned for that purpose. This means EC2 instances don't have to be overprovisioned to allocate the EBS volumes required. This architecture thus brings down compute costs dramatically.

Ondat aggregates I/O operations from different volumes to the backend EBS volume/s

This maximizes its usage and keeps the performance of the overall system close to the EBS limit threshold, whether it is bandwidth or IOPS.

Ondat allows the use of ephemeral disks for those workloads that don't require EBS backend volumes

At the same time, it guarantees a replica count defined by the user. This ensures that the price per I/O operation in a local SSD is radically better.

Ondat doesn't restrict the performance of volumes based on size, known as IOPSxGB

Kubernetes workloads usually require many smaller size disks instead of a few large ones. In an IOPSxGB model, the performance requirement of those smaller-sized volumes often means that you have to provision larger volumes than needed. This can be a pricey choice. Another alternative is to use costly provisioned IOPS.

Scalability improvements with Ondat in AWS

Ondat volumes are designed for containerized workloads

Hundreds of volumes can be provisioned and used with minimum overheads, in a matter of seconds. EBS has limits, given its overhead on the network and compute resources per node. It has a limit of 25-39 EBS disks per node depending on the EC2 instance type.

Ondat uses the Kubernetes worker nodes to expand its disk capacity

It provisions more capacity simply by adding more nodes to the cluster.

More backend EBS disks can be provisioned and given to Ondat to expand the current capacity of the cluster

While volume provisioning time is critical for application workloads, adding EBS backend disks for Ondat to use can be done asynchronously with no impact on running workloads. That leverages strong capabilities from EBS and Ondat to achieve the best outcome.

AWS EFS

In addition to EBS volumes, users often take advantage of the shared file system capabilities that EFS provides.

Ondat offers Kube-native shared file system volumes designed for containerized workloads. Unlike EFS volumes, their performance is not restricted to their size or usage. They are dynamically provisioned on-demand and they leverage all the Ondat volume functionalities, such as high availability, encryption, or topology-aware placement.

It is conventional to use a large NFS share and split the data from different workloads in directories, even when using EFS. Ondat, however, is not restricted by the architectural limitations of the past. Ondat ReadWriteMany (RWX) volumes place their data on the worker nodes of the cluster while using Kubernetes native objects (such as ClusterIP Services) to expose the shared file system endpoints. That keeps the data self-contained in a Kubernetes cluster and keeps the data per workload isolated. Every volume has its own user-space running NFS server dynamically provisioned with high availability.

Ondat is agnostic, hence the use of the RWX volumes is transparent regardless of the Kubernetes distribution or underlying Linux distribution OS.

Ondat and AWS

Choosing the right architecture to run scalable stateful workloads does not need to be hard. Using Ondat in your AWS architecture can reduce costs, improve performance and increase reliability, all by building upon the AWS services that you are using today.

Sign up for the Tech Preview

Learn how Ondat can help you scale persistent workloads on Kubernetes.

Get started for free