Organisations are starting to realise that leveraging artificial intelligence and machine learning, as a business tool, is important for success.
Increasingly, organisations are looking to automate data management operations, via AI and machine learning. They no longer view this practice as a risk, “but a leading-edge benefit,” according to Hari Mankude, CTO, Imanis Data.
AI tool helps marketers get to know their customers
Solving the data management conundrum
“Put into the context of a specific use-case, one of the challenges in data management is determining optimal RPO and RTO to support the business and then ensuring that these business requirements are met,” explains Mankude.
“This is where an enterprise data management platform can gather sufficient information from the backups and restores that have been occurring, and feed this into an AI engine to predict the RPO and RTO of a workflow,” he continues.
“The predictions from the AI engine can be inserted into a positive feedback loop to build auto learning policies that can dynamically alter the backup frequencies such that customer’s RPO requirements are met.”
Taking the next step with AI adoption — overcoming the problem of data
Essentially, because of the transformative nature of AI as a business tool, the customer will no be longer required to setup up complex policies for tasks, such as backup or replication.
Instead, Mankude suggests that a business requirement, such as Recovery Point Objective (RPO) or Recovery Time Objective, is specified and the data management platform leverages an AI engine to auto compute the dynamic policies.
AI: From academia into reality
“Being on the cutting edge of AI has given Imanis Data a front row seat to the evolutionary nature of how AI moved from academia into a real effector of business change,” says Mankude.
“A great example of this is how we developed our ThreatSensetm capability. One of the requirements that our customers surfaced was the ability to predict the change rate of their tables and databases. This change rate could vary based on the hour of the day, the day of the week, or based on other factors such as end of month applications, etc. At Imanis Data, we were able to address this requirement using ML, but we also discovered an interesting byproduct of satisfying this requirement.”
“In the process of building intelligence to “learn” the change rate, we realised that with our ML model, we could flag anomalies in the rate of change. One of the possibilities for these anomalies in the rate of change could be a ransomware attack on data on the primary cluster by encrypting the data. We subsequently realised that an ML model could be built to flag this anomaly in real time to the customer. This was the genesis of Imanis Data ThreatSensetm.”