Guest: Asaf Yigal (LinkedIn)
Company: Logz.io (Twitter)
Show: Let’s Talk

The amount of data that people are collecting today is a systematic challenge not only from a technological point of view but a process one too. Many organizations are still just lumping everything together and hoping to make sense of it but this can prove costly and inaccurate. However, observability company Logz.io believes that organizations need specific guidelines or objectives in place for observability and then to develop a system that is optimized for cost and mean time to resolution. Observability is supposed to make life easier for organizations but many are still spending too much time setting it up, filtering out the noise, and grappling with alert fatigue.

Over 50% of the alerts that are received result in no action taken, therefore it is crucial to reduce the alerts and filter out the unnecessary ones. While AI is being used in anomaly detection, the reality is that everything can potentially be seen as an anomaly whether it is a customer joining a cluster or a pod being restarted. Logz.io believes the way around this is to first define your SLOs and then set up the AI and anomaly detection.

There are two different sets of teams looking at the production environment which need to be considered, the engineering team who are focused on building an application to deliver a service for their customers. On the other hand, the DevOps/production engineering team is focused on allocating the infrastructure and ensuring the applications are laid out properly on the infrastructure from a performance, security, availability, and cost perspective. They are looking at two different sides of the same coin and observability needs to reflect this.

Generative AI is also proving useful for organizations but it is important to understand the limitations of it, such as the length of time needed to train the models and its reliance on old data. However, it can process a lot of information and provide the right answers that organizations need. Logz.io is using generative AI to help customers achieve their tasks more simply and quickly, being able to search in English rather than SQL or another language. It can also help identify and create sophisticated alerts.

Generative AI does carry with it some challenges, and one of the hot topics being discussed is intellectual property. It calls into question whose intellectual property the code is and who is responsible for it if you are using generative AI to write the code. Another key concern is accuracy and while it can be good from a creativity perspective when the AI will generate new ideas, there are also cases where inaccuracy can have negative consequences. Although many people have been concerned about AI taking their mundane jobs, the reality is it may replace some of the more sophisticated work relating to creativity.

This summary was written by Emily Nicholls.