Operations | Monitoring | ITSM | DevOps | Cloud

In large-scale AI model training, network performance is no longer a supporting actor — it’s center stage. Job Completion Time (JCT), the key metric for measuring training efficiency, is heavily influenced by the network interconnecting thousands of GPUs. In this post, learn why JCT matters, how microbursts and GPU synchronization delays inflate it, and how platforms like Kentik give network engineers the visibility and intelligence they need to keep training jobs on schedule.

In the push to simplify IT management, the idea of a single, all-encompassing AIOps platform is certainly appealing. The promise of one dashboard to monitor the entire IT stack—from applications and infrastructure to the network—suggests a world of streamlined operations. This generalist approach aims to provide a broad overview, correlating data from across the business to spot trends and potential issues.

A network map is a visual representation of the devices and connections that make up an IT network. For IT professionals, network maps are essential tools for monitoring performance, troubleshooting issues, enhancing security and planning infrastructure upgrades. There are multiple types of network maps, each serving a specific purpose, ranging from physical layout diagrams to cloud-based and security-oriented architectures.

A global airline experienced persistent Google Meet connectivity issues with no apparent network infrastructure faults. While their APM tool offered visibility into network paths, it didn’t surface any local anomalies. Catchpoint’s endpoint monitoring, however, revealed performance degradation specifically on Wi-Fi Channel 44 (5GHz band), where signal strength dropped to -80 dBm compared to optimal ranges of -30 to -50 dBm.

Kubernetes networking is deceptively simple on the surface, until it breaks, silently leaks data, or opens the door to a full-cluster compromise. As modern workloads become more distributed and ephemeral, traditional logging and metrics just can’t keep up with the complexity of cloud-native traffic flows.