Can you describe a specific instance where you implemented SRE practices to improve system reliability?