Automate stateful application reconstruction
Every disaster recovery plan is perfect. Until production actually fails.
As SaaS companies grow, their systems become increasingly complex. With larger Kubernetes environments and faster deployments, recovery becomes harder to manage.
Recovery steps are often manual, inconsistent and difficult to test. They depend on a few experienced engineers, fragmented tooling and fragile automation scripts.
What recovery actually requires:
Capture systems as they run.
Track how they change over time.
Understand how everything is interdependent.
Reconstruct everything in the right order.
Prove it works or at least know what’s wrong if it doesn’t.
Phantom - What is it?
Phantom Data Protection allow you to create twin instances of your applications. Twin applications can be used for testing, debugging, staging and recovery. Phantom manages the full lifecycle of twin applications including automatic destruction and cleanup.
The platform helps your organisation move faster by focusing on reducing:
How does it work?
In the Phantom model, applications are composed of components (e.g. an API housed in a K8s pod). Components live in runtimes (e.g. K8s cluster) and an environment is a collection of runtimes. Applications may span heterogeneous runtimes. Phantom continuously captures dependencies, state and topology.
Phantom reconstructs application components into isolated twin applications. It coordinates runtime and dependency ordering and reconstructs stateful systems using runtime-native semantics and verifies operational readiness. Twin applications and components can be validated, tested, promoted or discarded.
Infrastructure, storage, and security systems provide the supporting layer. That layer is responsible for runtime agents, object storage and encryption. It's also responsible for reconstruction-metadata and orchestration-state.
Phantom currently supports:
Reconstruction is coordinated using application topology and runtime relationships. Dependency sequencing and consistency metadata is also used. Reconstruction is deterministic.
We are a small team with more than 50 years of combined software industry expertise.
We have seen many organisations struggle with the complexities of distributed systems. “Frankly, I don’t know how we can recover this application after an incident”.
Glue code keeping tools and processes together can work for small teams. Fast moving and growing teams will need a different approach.
We believe the best tooling is operational.
Tooling that automates the tedious, error prone and boring tasks.
Tooling that lets software teams move faster with confidence. Tooling that eats the complexity.
That is what we are building.
Ville Suonurmi, founder, ville@phantomdp.com
Ville has nearly 30 years of software industry experience. He has worked with infrstruture software at IBM, Oracle, HPE Software and Citrix. He also studies software engineering at the University of Helsinki.