Machine learning data management company Imanis Data has introduced an autonomous backup product powered by machine learning.
The firm said users can specify a desired RPO (Recovery Point Objective) and its SmartPolicies tech then set up the backup schedules. The tech is delivered as an upgrade to the Imanis Data Management Platform (IDMP) product.
SmartPolicies uses metrics including criticality and volume of data to be protected, primary cluster workloads, and daily or seasonal resource utilisation, to determine the most efficient way to achieve the desired RPO.
If it can't be met because, for example, production systems are too busy, or computing resources are insufficient, then SmartPolicies provides recommendations to make the RPO executable.
Other items in the upgrade include any-point-in-time recovery for multiple NoSQL databases, better ransomware prevention and general data management improvements, such as job tag listing and a browsable catalog for simpler recovery.
Imanis changed its name from Talena in July last year and told us at the time: "We occasionally got confused for doing 'talent management' or sometimes people heard the word 'Talend' and assumed that."
The firm rounded up a canned quote from Christophe Bertrand, senior analyst for data protection at ESG Research, who said: "The data protection market in this space is underserved by traditional vendors and Imanis Data with their unique machine learning approach is setting the bar for Hadoop and NoSQL enterprise data management."
Supported Hadoop and NoSQL platforms include Apache Cassandra, Apache HBASE, Cloudera, Couchbase, DataStax, Hadoop, HDInsight, HortonWorks, Microsoft ADLS, MongoDB, and Vertica platforms. Additionally, the latest release includes usability enhancements to Imanis Data FastFind including job tag listing and a browsable catalog for simple recovery.
Imanis said the anti-ransomware addition, ThreatSense, has so-called Intelligence Augmentation, which allows users to report false positives only identifiable through human observation – this data is fed back into the machine learning model, eliminating specific "anomalies" and making its ransomware detection more effective.
Having backup software set up its own schedules based on input RPO values isn't a new idea, but having it done with machine learning is. The checking of available resources is a darn good idea too and, when you think about it, absolutely necessary.
Otherwise "backup run failed" messages would start popping up all over the place – not good. We expect other backup suppliers to follow in Imanis's wake and start sporting "machine learning-driven policy" messages quite quickly. ®