Rok Data Management Platform Overview
Rok is the global data distribution layer that makes your apps truly performant and at the same time truly portable, in a fraction of the cost of a shared storage solution. The Rok data management and storage platform gives you the best of both worlds: the performance of local storage and the flexibility of shared storage.
Rok allows you to run your stateful containers over fast, local NVMe storage on-prem or on the cloud, and still be able to snapshot the whole application, along with its data and distribute it efficiently: across machines of the same Kubernetes cluster, or across distinct locations and administrative domains over a decentralized network. This hybrid approach to container networking makes data management for machine learning and real-time big data applications simple, without sacrificing performance.
Rok Cluster
Your applications may run over local NVMe SSDs, but you keep the flexibility feature, as if they were running over shared storage. Rok provides an enterprise data management layer that makes it possible for you to instantly snapshot your containers for local and offsite backups. Take immutable, group consistent snapshots of your apps and keep these snapshots in a backup store, e.g. S3. Combined with Rok’s instant cloning, you can recover from hardware failure in minutes, regardless of the volume’s capacity.
Rok Registry
Rok Registry is the single pane of glass where you search, discover, and share your datasets and environments with other users. Users can create private or public groups on Rok Registry and can define fine-grained Access Control Lists. The Registry cloud data management platform gives you full control of your data over individual users, locations, and devices, thus ensuring your sensitive data remains secure.
No vendor lock-in. With Rok’s cross platform data management software you can use any type of hardware for compute and storage. You can also move to the cloud provider of your choice, or even move among different cloud providers.
How our Data Management Platform Works
Create A Local Bucket
Take instant snapshots of your containers locally. Group them into collections. We call these collections Buckets.
Publish A Local Bucket
Create a link to your Bucket on Rok Registry.
Set Permissions
You can share your Bucket links with specific collaborators, groups of collaborators, everyone, or just yourself! As far as everyone else is concerned your Bucket does not exist.
Search For A Bucket
Search on Rok Registry for the Bucket you want to sync and copy its link.
Subscribe
Go to your Rok and create a new subscribed Bucket by pasting the link you got from Rok Registry.
Let The Syncing Begin!
Subscribers exchange data among them in a decentralized fashion. As the publisher updates their Bucket with new snapshots, the subscribers automatically sync them.
Spawn Your Snapshots
Once a snapshot is synced, you are ready to instantly spawn a container from this snapshot on the target cluster.
Free Technical Workshop
Turbocharge your team’s Kubeflow and MLOps skills with a free workshop.