BigPanda announces Generative AI capabilities

Written By:
Content Copyright © 2023 Bloor. All Rights Reserved.
Also posted on: Bloor blogs

BigPanda announces Generative AI capabilities banner

It looks like it has been a busy and interesting few months for AIOps vendor BigPanda, that might well be pivotal for it in particular, but also, more generally, for the whole AIOps market.

The last time I spoke with Blair Sibille, Field CTO at BigPanda, back in March, it was obvious that significant developments were afoot. While event correlation and automation remained at the heart of the Big Panda offering, Blair talked passionately about the way it was re-architecting its whole approach to data and the potential impact Generative AI and Large Language Models (LLMs) on the challenge of making sense of the tsunami of event data that IT operations teams face, and the tantalising potential of genuinely predicting outages and performance degradation before users reported them. However, he was not ready to go public, so I held back from writing anything up.

Clearly, there is no longer the same reticence. In May Blair posted an interesting blog on the results of experiments the company ran on real life observability and monitoring data in three scenarios using LLMs. I’ll let you read the article. It shows clearly how LLMs can help identify the root cause of a problem automatically although, as my colleague David Norfolk remarked “BigPanda had hindsight. They knew exactly what the problem was. And the initial scenario that it was disappointed with could have been equally plausible as the others without the benefit of hindsight.” There is obviously real potential for LLMs to speed up and improve the process of identifying root cause analysis of IT problems. But perhaps, like early automobiles, we still need someone to walk in front waving a flag.

On 11th July BigPanda announced its Generative AI capabilities to the world. So, it clearly has confidence in Generative AI, although I note that it states that it “suggests root causes”. So, still a need for the person with the flag. A blog from Blair Sibille on the same day provides a lot more detail on the Generative AI announcement. Sensibly, he goes into some detail to cover off concerns around security and data privacy as well as describing the rationale behind the move to Generative AI and diving a little more deeply into what it does.

This brings me on to a 19th July BigPanda webcast of a discussion between Jason Walker, Chief Technology Officer at BigPanda and Nick Heudecker, Senior Director of Market Strategy and Competitive Intelligence at Cribl, talking about what the integration of the capabilities between the two companies means for getting stronger actionable insights from all of that observability data being generated. Big Panda ingests data from lots of different vendors tools.

So, on the face of it, why is this any different? The answer lies in the way that Cribl’s flagship product, Cribl Stream enables you to filter, shape and enrich data before it gets into the data stream. It captures all the MELT data (Metrics, Events, Logs, Traces) and the full fidelity content can be delivered to a low-cost storage environment for later analysis, if needed. In the meantime, that data has been whittled down and enhanced to deliver a smaller stream of data that reduces “noise” and amplifies the “signal”. The result is more relevant data for BigPanda’s AI to work on, getting faster responses and significantly reduced data transmission and storage costs. Both Jason and Nick saw this “shift-left” movement of data filtering and enrichment as making the goal of predicting rather than reacting to infrastructure issues a reality rather than an aspiration. For me, it also made sense of the work BigPanda has been doing to reposition itself as much about organising data as about event correlation. Hopefully, I will be getting a brief from BigPanda later this month at which I can ask a lot more questions and report back.

Ultimately, these developments have implications for the whole cloud and hybrid IT infrastructure management market. Having some ML (machine learning) capabilities or algorithms doesn’t, in and of themselves, give you AIOps solutions. And some existing data architectures are no longer going to be able to handle the vast amount of raw data (and noise) being generated. That AIOps is now deemed as a very important component of the whole cloud management space, has possibly been highlighted by Dell’s announcement of their intention to acquire Moogsoft, one of the AIOps pioneers. I’ll leave any detailed analysis of that move for when the dust settles. But Dell must feel that AIOps has become a critical component in helping its customers deliver business service assurance.

Post a public comment?