Our hope is that we have perhaps distanced ourselves away from being specific to Cloud Foundry. There’s a lot of neat things happening everywhere – for example Kubernetes – and what we don’t want is to have the system be so opinionated that it would only be useful within Cloud Foundry.

Loggregator is doing its best to be a distributed logging and metrics system for several distributed systems, and not just Cloud Foundry. We had our original iteration where everything was specific to Cloud Foundry; every metric type was something that only Cloud Foundry could ever care about, and the consumers that would only ever care were at Cloud Foundry. What we noticed was other interesting distributed systems were coming up, such as Kubernetes, and Loggregator was in a technical position where it could service these things, but we were so opinionated about Cloud Foundry we really worked putting ourselves in front of that.

We decided to move our API more towards a generic place, and have generic metrics with a subset of the information we previously had, that would be useful therefore for any different number of distributed applications, and enable distributed applications to use Loggregator to optimize their apps, or maintain them, do DevOps work and just ensure that iterations are happening.

Pivotal’s main focus is to enable to do interesting things, so Cloud Foundry wants to enable to do hard, complex distributed systems, and maintain them, and iterate on their software quickly, and have confidence that their stuff is actually working, so they need visibility in that.

Source link


Please enter your comment!
Please enter your name here