https://www.mfun.ink/en/categories/aiinfra/ Recent content in AIInfra on Mengboy Tech Notes Hugo -- 0.156.0 en Wed, 01 Apr 2026 01:19:20 +0000 https://www.mfun.ink/en/2026/04/01/claude-openai-dual-stack-degrade-runbook/ Wed, 01 Apr 2026 01:19:20 +0000 https://www.mfun.ink/en/2026/04/01/claude-openai-dual-stack-degrade-runbook/ <p>Running both Claude and OpenAI in production sounds resilient—until a <strong>slow failure</strong> hits: latency climbs, 429s spike, quality drifts, and everything still looks “up.”</p> <p>This guide gives you a practical dual-stack degradation runbook: timeout probing first, circuit-based cutover second, and an error-budget dashboard to keep business impact bounded.</p> https://www.mfun.ink/en/2026/03/30/claude-openai-dual-provider-gateway-failover-sla/ Mon, 30 Mar 2026 01:14:00 +0000 https://www.mfun.ink/en/2026/03/30/claude-openai-dual-provider-gateway-failover-sla/ <p>If your production stack calls both Claude and OpenAI, the hard part is not API integration. The hard part is keeping user experience stable when one provider starts throwing 429/5xx spikes, regional latency, or timeout storms.</p> <p>This guide gives you a practical dual-provider gateway playbook: health probes, circuit breaking, SLA-aware fallback, and observability loops. The goal is not “never fail.” The goal is <strong>controlled failure with controlled cost and controlled latency</strong>.</p>