--- title: "To make a service more stable, eliminate dependencies" url: https://daily.dev/posts/to-make-a-service-more-stable-eliminate-dependencies-syibv7oj8 source_url: https://itnext.io/to-make-a-service-more-stable-eliminate-dependencies-d60d1c1a65bc type: article source: "ITNEXT" published: 2026-08-15T22:35:21.190Z updated: 2026-08-15T22:35:41.846Z tags: ["career", "observability", "distributed-systems", "envoy"] reading_time: 3 upvotes: 0 comments: 0 language: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # To make a service more stable, eliminate dependencies **[ITNEXT](https://daily.dev/sources/itnext)** · 3 min read · 0 upvotes · 0 comments ## Summary Every dependency a service relies on is another way for it to fail, so reliability engineering means questioning which dependencies are truly necessary and designing graceful degradation for the rest. Edge systems like load balancers and API gateways tend to be dependency-light because they carry outsized responsibility for platform availability. Envoy Proxy is offered as an example: it fetches xDS configuration periodically and caches it in memory, so request processing continues even if the xDS service goes down. Observability (logging, tracing) should be treated as best-effort rather than a hard dependency, using async logging and buffering so customer traffic isn't blocked by observability outages. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does Envoy Proxy stay available if the xDS control plane goes down? Envoy continues routing traffic using its last known configuration. It fetches configuration from the xDS service periodically and stores it in memory, so request processing relies on that cached local state rather than calling xDS for every request. This means an xDS outage does not stop traffic from being routed, even though Envoy still depends on xDS for updates. _Anyone designing resilient proxy or gateway systems can track patterns like this on daily.dev._ ### Should logging and tracing be treated as hard dependencies in a production service? No, observability should be best effort rather than a hard requirement for serving traffic. If a logging or tracing backend becomes unavailable, customer traffic should keep flowing rather than stopping. This is achieved through asynchronous logging, buffering, and truncation policies so operational visibility issues never block customer-facing availability. _Engineers weighing observability tradeoffs against uptime can follow this reasoning on daily.dev._ ## Similar posts on daily.dev - [The closer to the edge, the more stable a platform must be](https://daily.dev/posts/the-closer-to-the-edge-the-more-stable-a-platform-must-be-1rptuv5sk) · ITNEXT · 0 upvotes · 0 comments - [Reliability is a Product Decision](https://daily.dev/posts/reliability-is-a-product-decision-wpmua7ypy) · Erlang Solutions · 1 upvotes · 0 comments --- Tags: [#career](https://daily.dev/tags/career), [#observability](https://daily.dev/tags/observability), [#distributed-systems](https://daily.dev/tags/distributed-systems), [#envoy](https://daily.dev/tags/envoy) [View this post on daily.dev](https://daily.dev/posts/to-make-a-service-more-stable-eliminate-dependencies-syibv7oj8)