Paige Cruz

@paigerduty.com

201 Followers 390 Following

🔮 future alpaca shepherdess ➡️ currently education eng + CNCF Ambassador ⚙️ former SRE blog: https://www.paigerduty.com/

simple t-shirt made of knit stitches with 2 by 2 rib along the hems
grainy photo of the 1984 IBM M2 keyboard from ebay
Paige giving a thumbs up in a fuzzy blue gopher hoodie - the hood is the head of Goldie
So, What Are Agents and Collectors? If a telemetry pipeline's job is to collect, process and export telemetry data, then what are agents and collectors for? If your head is already spinning, trust me, I've been there. When | first started exploring telemetry pipelines, | wanted to define clear, universal distinctions between agents, collectors and telemetry pipelines. But the deeper | dug, the more | realized that these terms don't always map neatly to strict definitions — different teams and vendors use them in slightly different ways. That said, it's still helpful to have a general frame of reference. Agents: The First Mile of Telemetry Agents are like your local mail carrier — they operate at the source, collecting telemetry from a specific application or system and forwarding it onward to a central location. They typically run alongside applications as sidecars or on the 'same host, focusing on that “first mile" of data delivery. Collectors: The Middle Mile Aggregators Collectors are like a regional post office, gathering telemetry data from multiple agents (or directly from applications and infrastructure) before forwarding it to its final destination. Collectors can receive and route telemetry. Telemetry Pipelines: The Entire System A telemetry pipeline is like the entire postal system and is responsible for handling observabilty data end-to-end. It does not just forward data but can normalize, enrich, filter and dynamically route telemetry based on your needs.
Picture of my desk set up, a wall full of surreal art prints. The desk is cluttered but not totally taken over by tech, wires, lights, and notebooks.
Norman a sleek black cat is curled up into a ball with his tail tucked in clearly in deep rest and relaxation
Hanson "We work with crashes. When you talk about production data, it really is crashes. Then you go on Android and say 'What's the p50 in the last 24 hours?' 'Great! Tell me more!' 'Nahhh'
A very good but not perfect circle drawn onscreen with a score of 96.3% from neal.fun
3 colors of linen squares in a checkerboard quilted together with thick cotton thread in a grid
paige chomping on a small wooden block engraved 2 years