Lead Data Engineer | Data Architecture & DevOps | 10+ ans ร builder (et casser) des pipelines data qui scalent | Je partage mes war stories en prod pour que tu dormes mieux ๐ด
๐ ๐๐ฎ๐ป ๐๐ผ๐ ๐ฒ๐ ๐ฝ๐น๐ฎ๐ถ๐ป ๐ฅ๐ฃ๐ข, ๐ฅ๐ง๐ข, ๐ช๐ฅ๐ง, ๐ฎ๐ป๐ฑ ๐ ๐ง๐ ๐ถ๐ป ๐น๐ฒ๐๐ ๐๐ต๐ฎ๐ป ๐๐ต๐ฟ๐ฒ๐ฒ ๐๐ฒ๐ป๐๐ฒ๐ป๐ฐ๐ฒ๐?
๐๐ฆ๐ช๐ต๐ฉ๐ฆ๐ณ ๐ค๐ข๐ฏ ๐. ๐๐๐ ๐ฎ ๐ฑ๐ถ๐ฎ๐ด๐ฟ๐ฎ๐บ ๐ฐ๐ฎ๐ป. ๐
๐ The ๐ง๐๐จ๐๐ก๐๐๐ฃ๐๐ฎ ๐ค๐ ๐ฅ๐ง๐ค๐๐ช๐๐ฉ๐๐ค๐ฃ ๐จ๐๐ง๐ซ๐๐๐๐จ is rooted in the ๐ฆ๐๐ (Service Level Agreement), which defines the commitments your services must meet.
๐ The ๐ค๐ผ๐ฆ (Quality of Service) ensures these commitments are maintained by tracking critical metrics such as ๐ฅ๐ฃ๐ข, ๐ฅ๐ง๐ข, ๐ช๐ฅ๐ง, and ๐ ๐ง๐, which are essential for ensuring resilience and effective recovery from disruptions.
๐ ๐ง๐ต๐ฒ ๐ฐ๐ผ๐ป๐ฐ๐ฒ๐ฝ๐๐, ๐บ๐ฎ๐ฑ๐ฒ ๐๐ถ๐บ๐ฝ๐น๐ฒ:
โก๏ธ ๐ฅ๐ฃ๐ข: How much data can you afford to lose? (e.g., backups every 15 minutes = 15 min max data loss)
โก๏ธ ๐ฅ๐ง๐ข: How long can you afford to be down? (e.g., restore operations within 6 hours)
โก๏ธ ๐ช๐ฅ๐ง: How long to verify everything is really fixed? (e.g., application, database, and log checks)
โก๏ธ ๐ ๐ง๐: The total time you can survive disruption (RTO + WRT = MTD).
๐ก ๐๐ฒ๐ฒ๐ธ๐ ๐ฎ๐ป๐ฎ๐น๐ผ๐ด๐: Think of it like this:
- ๐ฅ๐ฃ๐ข is "๐ป๐๐ค ๐๐๐ ๐๐๐๐ ๐๐ ๐ก๐๐๐ ๐๐๐ ๐ผ ๐๐๐ค๐๐๐ ๐ ๐๐๐๐๐ฆ?"
- ๐ฅ๐ง๐ข is "๐ป๐๐ค ๐๐๐ ๐ก ๐๐๐ ๐ผ ๐๐๐ ๐ข๐๐ ๐กโ๐ ๐๐๐๐ ๐๐๐ก๐๐ ๐ ๐๐๐๐ โ?"
- ๐ช๐ฅ๐ง is "๐ท๐๐ ๐ผ ๐๐๐๐ ๐กโ๐ ๐๐๐โ๐ก ๐ ๐๐ฃ๐ ๐๐๐๐, ๐๐๐ ๐๐ ๐๐ก ๐ค๐๐๐๐๐๐?"
- ๐ ๐ง๐ is "๐บ๐๐๐ ๐๐ฃ๐๐ ๐๐ ๐ผ ๐๐๐โ๐ก ๐๐๐ฅ ๐๐ก ๐๐ ๐ก๐๐๐."
๐ฏ ๐๐น๐ถ๐ฐ๐ธ ๐ผ๐ป ๐๐ต๐ฒ ๐ฑ๐ถ๐ฎ๐ด๐ฟ๐ฎ๐บ ๐ณ๐ผ๐ฟ ๐ฎ ๐๐ต๐ผ๐ฟ๐๐ฐ๐๐ ๐ฏ๐ฎ๐ฐ๐ธ ๐๐ผ ๐ฐ๐น๐ฎ๐ฟ๐ถ๐๐.
Because letโs face it: sometimes, ๐ ๐ฅ๐๐๐ฉ๐ช๐ง๐ ๐๐ค๐๐จ๐ฃโ๐ฉ ๐๐ช๐จ๐ฉ ๐จ๐ฅ๐๐๐ ๐ ๐ฉ๐๐ค๐ช๐จ๐๐ฃ๐ ๐ฌ๐ค๐ง๐๐จโ๐๐ฉ ๐จ๐๐ซ๐๐จ ๐ ๐ฉ๐๐ค๐ช๐จ๐๐ฃ๐ ๐๐ค๐ช๐ง๐จ ๐ค๐ ๐๐ค๐ฌ๐ฃ๐ฉ๐๐ข๐.