1.
SQL Server Architecture
SQL Server has two main engines: Relational Engine and Storage Engine.
Relational Engine: Handles query parsing, optimization, execution plans.
Storage Engine: Handles data storage, reading/writing pages, transaction logging.
SQLOS: Manages memory, scheduling, I/O, latches, spinlocks.
Query lifecycle: Parsing → Optimization → Execution → Storage access.
2. System Databases
master: Stores metadata, logins, linked servers, configuration.
msdb: Stores jobs, backup history, log shipping details.
model: Template DB, copied whenever a new DB is created.
tempdb: Used for sorting, row versioning, temporary tables, version store.
Best practices: multiple data files, set growth properly, monitor usage.
3. Backup & Recovery
Recovery Models: Full, Bulk-Logged, Simple.
Backup types: Full, Differential, Transaction Log, Copy-Only.
Restore sequence: Full → Differential → Log backups.
Point-in-time recovery using STOPAT in log restore.
Tail-log backup is taken before restore in failure scenarios.
4. Performance Tuning
Indexes: Clustered, Non-Clustered, Columnstore, Filtered.
Index fragmentation: use REORGANIZE (minor) or REBUILD (major).
Statistics help optimizer with data distribution.
Locking, Blocking, Deadlocks: handled with DMVs.
Execution plans show actual/estimated cost of queries.
5. High Availability & Disaster Recovery
Log Shipping: Copy & restore logs between servers.
Database Mirroring: High safety (sync) or performance (async).
Always On AG: Multiple replicas, automatic failover.
Failover Cluster: Instance-level HA.
RPO = data loss tolerance, RTO = downtime tolerance.
6. Security
Logins (server) vs Users (database).
Server roles: sysadmin, securityadmin, dbcreator, etc.
Database roles: db_owner, db_datareader, db_datawriter.
Permissions: GRANT, DENY, REVOKE.
Fix orphaned users: ALTER USER WITH LOGIN.
7. Replication
Types: Snapshot, Transactional, Merge.
Publisher (source), Distributor (middle), Subscriber (destination).
Transactional replication: near real-time, uses log reader.
Merge replication: supports offline updates with conflict resolution.
8. Monitoring & Maintenance
Backups, Index maintenance, Integrity checks.
DBCC CHECKDB for corruption detection.
Monitor logs: sp_readerrorlog.
Agent Alerts & Notifications with Database Mail.
Custom monitoring: DMVs for blocking, performance.
9. Cloud (Azure SQL)
Azure SQL Database: PaaS, auto backups, limited control.
Managed Instance: Closer to on-prem SQL, supports Agent.
SQL VM: Full control, same as on-prem.
Geo-replication, failover groups for DR.
Security: AAD integration, firewall rules.
10. Common Interview Q&A;
Q: How do you troubleshoot slow queries?
A: Check execution plan, statistics, missing indexes, blocking.
Q: DB in suspect mode?
A: Check logs, run emergency mode repair, restore if needed.
Q: Blocking vs Deadlock?
A: Blocking = waiting, Deadlock = circular wait; resolved by killing one process.
Q: Move login between servers?
A: Use sp_help_revlogin to script SID and password.
