Angielski dla DBA — słownictwo administratora baz danych

Twoja baza obsługuje 10 milionów zapytań dziennie — ale na incident calu o 2 w nocy nie wiesz jak opisać problem po angielsku

Alarm o 2:47. Slack pełen czerwonych alertów. Zagraniczny team lead pisze: "Primary is unresponsive — what's the status?" Wiesz dokładnie co się dzieje i jak to naprawić — ale szukasz słów w języku, w którym musisz teraz działać.

Jeśli pracujesz jako DBA, Data Engineer, Database Developer, Cloud Database Engineer lub Backend Developer z dużą ekspozycją na SQL, ten artykuł daje Ci precyzyjny język do incydentów, performance review i capacity planningu.

Architektura baz danych — 14 terminów

Wydajność — 9 terminów

Operacje — 9 terminów

Scenariusze komunikacji

a) Incident response — baza nie odpowiada — 8 zwrotów

• "We're experiencing a critical outage — the primary database is unresponsive. Initiating failover to the replica now."
• "Failover completed successfully. The replica is now accepting writes. Estimated data loss is less than 30 seconds — within the RPO."
• "Root cause is a disk I/O spike caused by an unoptimised batch job running outside the maintenance window."
• "The primary database is back online. I'm monitoring replication lag before switching traffic back."
• "Connection pool is exhausted — increasing pool size as an immediate mitigation."
• "We have deadlocks on the payments table — I've identified the conflicting transactions and am applying a hotfix."
• "Replication lag is at 4 minutes and climbing — I'm pausing non-critical write traffic to let the replica catch up."
• "The incident has been resolved. Total duration: 18 minutes. A post-mortem will be published within 48 hours."

b) Performance review z deweloperami — 6 zwrotów

• "This query is causing a full table scan on a 50M-row table — we need a composite index on user_id and created_at."
• "The execution plan shows a nested loop join here — it's O(n²) for large datasets. Consider rewriting with a hash join."
• "Your ORM is generating N+1 queries — 847 round trips for a single page load. Add eager loading for related entities."
• "This transaction holds a table-level lock for 12 seconds — it's blocking all concurrent writes. Let's reduce the lock scope."
• "The stored procedure has an implicit cursor loop — processing rows one at a time. A set-based rewrite will be significantly faster."
• "I'd recommend running EXPLAIN ANALYSE before submitting any query that touches tables over 1 million rows."

c) Capacity planning — 6 zwrotów

• "Based on current growth projections, we'll hit storage limits in approximately 8 months."
• "CPU utilisation is trending at 78% during peak — we'll need to scale up or shard before the next major campaign."
• "I recommend archiving records older than 24 months — it'll reduce the primary table size by 60% and improve query performance immediately."
• "The connection pool ceiling of 200 will become a bottleneck at 3x current traffic — we should implement PgBouncer."
• "Write throughput is currently 4,200 TPS at peak — the current instance handles up to 8,000 before we see degradation."
• "The proposed data retention policy will reduce storage costs by approximately 40% annually."

Krótki dialog — DBA na incident calu z międzynarodowym zespołem o 2 w nocy

SQL i komunikacja z deweloperami — 8 fraz do code review

• "The migration adds a column without a default value on a live table — this will lock the table for the duration. Please add a default or use a deferred constraint."
• "The query is correct, but it'll be slow at scale. I've added an inline comment with a suggested index — please create it before this goes to production."
• "We need to review the database changes in staging first. Can you add migration tests to the PR?"
• "This stored procedure has no error handling — a failed INSERT will silently corrupt the audit log. Please wrap it in a TRY/CATCH block."
• "The EXPLAIN output shows an index scan costing 12,000 units — it should be under 1,000. Can we discuss before it merges?"
• "I'd strongly recommend wrapping these two operations in a single transaction. If step 2 fails, step 1 leaves the database in an inconsistent state."
• "The ORM is issuing a SELECT * — in production this table has 140 columns. Please specify the columns you actually need."
• "This could cause a deadlock with the existing nightly batch job. Let me check the lock acquisition order before we deploy."

Najczęstsze błędy Polaków

1. "the base" → "the database". Base nie istnieje w angielskim jako skrót od database. ❌ "There's a problem with the base." → ✅ "There's a problem with the database." / ✅ "The DB is unresponsive."

2. "query is slow" → zbyt ogólne. ❌ "This query is slow." → ✅ "This query is causing a full table scan and taking 8 seconds on a 50M-row table."

3. "I will make a backup" → nieprawidłowe kolokacje. ❌ "I will make a backup." → ✅ "I'll take / run / create a backup."

4. "the table is locking" → nieprecyzyjna terminologia. ❌ "The table is locking." → ✅ "We have lock contention on the payments table." / ✅ "The table is locked by a long-running transaction."

5. "delete the indexes" vs "drop the indexes". W SQL: DROP dotyczy obiektów (tabele, indeksy), DELETE dotyczy wierszy danych. ❌ "Please delete the unused indexes." → ✅ "Please drop the unused indexes."

Quick Reference Table

Podsumowanie

DBA w środowisku międzynarodowym to nie tylko administrator bazy — to osoba, która tłumaczy problemy techniczne na język zrozumiały dla deweloperów, managerów i partnerów zewnętrznych. Precyzja terminologii w incydencie, code review i capacity planningu bezpośrednio wpływa na czas rozwiązania problemu.

Więcej kontekstu IT znajdziesz w artykule o słownictwie IT po angielsku. Jeśli pracujesz blisko zespołu analitycznego, przydatny będzie też artykuł o angielskim dla analityka biznesowego IT.

Gotowe fiszki z terminologią DBA znajdziesz w ścieżce Database Administrator w sekcji IT & Programowanie.