I recently had the opportunity to speak with Peter Smails and Jay Desai from Imanis Data. They provided me with an overview of what the company does and a view of their latest product announcement. I thought I’d share some of it here as I found it pretty interesting.
A big part of the Imanis Data story revolves around the “three pillars” of data management, namely:
- Protection – providing redundancy in case of a disaster;
- Orchestration – moving data around for different use cases (eg. test and dev, cloud migration, archival); and
- Automation – using machine learning to automate the data management functions, eg. Detecting anomalies (ThreatSense), SmartPolicies for backups based on RPO/RTO
The software itself is hardware-agnostic, and can run on any virtual, physical, or container-based platform. It can also runs on any cloud, and hence on any storage. You start with 3 nodes, and scale out from there. Imanis Data tell me that everything runs in parallel, and it’s agentless, using native APIs for the platforms. This is a big plus when it comes to protecting these kinds of workloads, as there’s usually a large number of hosts involved, and managing agents everywhere is a real pain.
It also delivers storage optimisation services, and supports erasure coding, compression, and content-aware deduplication. There’s a nice paper on the architecture that you can grab from here.
So what’s new with 4.0?
Any Point-in-time Recovery
Imanis Data now provides APITR for Couchbase, MongoDB, & Cassandra
- APITR can be enabled at bucket level for Couchbase;
- APITR can be enabled at repository level for Cassandra and MongoDB;
- Aggressively collects transaction information from primary database; and
- At time of recovery, user can pick a date & time.
ThreatSense “learns” from human input and updates the anomaly model. It’s a smart way of doing malware and ransomware detection.
- Autonomous RPO-based backup powered by machine learning;
- Machine learning model built based on cluster workloads and utilisation;
- Model determines backup frequency & resource prioritisation;
- Continuously adapts to meet required RPO; and
- Provides guidance on required resources to achieve desired RPOs.
I do a lot with a number of data protection vendors in various on-premises and cloud incantations, but I’m the first to admit that my experience with protection mechanisms for things like NoSQL is non-existent. It seems like that’s not an uncommon problem, and Imanis Data has spent the last 5 or so years working on fixing that for folks.
I’m intrigued by the idea that policies could be applied to objects based on criteria beyond a standard RPO requirement. In the enterprise I frequently run into situations where the RPO is often at odds with the capabilities of the protection system, or clashing with some critical processing activity that happens at a certain time each night. Getting the balance right can be challenging at the best of times. Like most things related to automation, if the system can do what I need it to do in the time I need it to happen, I’m going to be happy. Particularly if I don’t need to do anything after I’ve set it to run.
Imanis Data seems to be offering up a pretty cool solution that scales well and does a lot of things that are important for protecting critical workloads. Imanis Data tell me they’re not interested in the relational side of things, and are continuing to focus on their core competency for the moment. It looks like pretty neat stuff and I’m looking forward to see what they come up with in the future.