IBM buys Databand to keep bad data from tainting your AI
Upstart promises to tackle that old adage: Garbage in, garbage out
IBM says its acquisition of observability startup Databand will help it stop bad data from poisoning customers' machine-learning and analytics workloads and save countless hours of lost time in the process.
Founded in 2018, Israel-based Databand specializes in detecting and resolving data-related snafus by catching process and pipeline errors before they cause problems further up the stack.
The basic idea here is that while things like machine learning are a powerful tool for divining trends within complex systems or datasets, they’re reliant on a steady supply of quality data. As the old adage goes “garbage in, garbage out.”
Big Blue is positioning Databand as the sieve that’s going to catch missing or malformed data that would otherwise throw a wrench in the works and render unreliable results.
“You can’t protect what you can’t see, and when the data platform is ineffective, everyone is impacted, including customers,” Databand CEO Josh Benamram said in a statement.
Databand is the latest in a lengthy string of mergers and acquisitions following Arvind Krishna’s takeover as IBM CEO in 2020. During his tenure, the IT giant has acquired more than two dozen companies to bolster various aspects of its expansive portfolio.
- Oracle, IBM losing ground to local databases in China, says IDC
- IBM wins National Savings & Investment bank tech and ops deal
- Actual quantum computers don't exist yet. The cryptography to defeat them may already be here
- IBM's first cloudy mainframes scheduled to launch this week
“With the addition of Databand, IBM offers the most comprehensive set of observability capabilities for IT across applications, data, and machine learning,” Daniel Hernandez, GM for Data and AI at Big Blue, said in a canned statement.
Databand’s observability suite — which integrates with more than two dozen software platforms, including the major cloud providers, database services, and data analytics platforms — slots in alongside IBM’s Watson Studio data science suite, and its Instana application performance monitoring (APM) platform.
IBM acquired Instana in late 2020, with a similar goal of improving the recommendations offered by its Watson AI suite.
While Instana targeted application discovery and monitoring for workloads running across multiple cloud and on-premises environments, Databand is designed to make sure that data stays intact.
Once deployed, Databand searches for anomalous patterns in data streams and alerts engineers if any are found. Meanwhile, Instana provides customers a means to identify the source of that erroneous data and accelerate remediation of those services.
Together, IBM claims the two platforms provide a comprehensive view of their application and data.
While neither party disclosed the terms of the deal, IBM said it would retain all Databand employees and continue offering the service both as a subscription-based on-prem deployment or as a software-as-a-service application. ®