Module 1 · Orientation and Load Balancing · Lesson 2 of 18

Quick cheat sheet

3 min readTechInterviewNotes Editorial TeamUpdated 17 Sept 2026Save

Topic

Topic	Solves	.NET / C# hook	Main trap
Load balancers	Spread traffic, fail over, enable deploys	Map health/readiness endpoints; handle forwarded headers	Bad health checks and sticky state
Caching	Reduce latency and DB/API load	`IMemoryCache`, `IDistributedCache`, cache-aside	Stale data, stampedes, data leaks
SQL vs NoSQL	Choose persistence model	EF Core for SQL; abstractions for document/key stores	Picking by hype instead of access pattern
Indexing	Speed selective reads and sorting	`HasIndex`, unique indexes, query-plan review	Over-indexing and wrong composite order
Sharding	Scale writes/storage beyond one DB	Tenant-aware `DbContext` factory/shard resolver	Cross-shard queries and rebalancing
Replication	Read scale, HA, DR	Primary for writes, replicas for stale-ok reads	Replica lag and failover ambiguity
Message queues	Async work and spike smoothing	Outbox table + `BackgroundService` publisher	Duplicate messages and poison retries
Rate limiting	Protect shared capacity	`AddRateLimiter` partitioned by tenant/user/key	Wrong identity and retry storms
CDN	Edge cache and origin protection	`Cache-Control`, `ETag`, static asset versioning	Caching private data
Consistency	Define read/write guarantees	EF transactions, `rowversion`, optimistic concurrency	Mixing caches/replicas without rules
Eventual consistency	Scale cross-service workflows	Versioned events, projectors, reconciliation	Promising immediate consistency
Idempotency	Make retries safe	Idempotency-Key table; processed message IDs	In-memory key stores and body mismatch
Observability	Understand production behavior	`ILogger`, `ActivitySource`, `Meter`, health checks	PII leakage and high-cardinality metrics
AuthN/AuthZ	Identity and permissions	JWT bearer, policies, resource checks	Trusting client tenant/user claims
API design	Stable contracts	Minimal APIs, `ProblemDetails`, OpenAPI metadata	Leaking entities and weak pagination
Multi-tenant	Serve many customers safely	Tenant context, query filters, per-tenant quotas	Isolation leaks and noisy neighbors
Background jobs	Run slow/scheduled work	`BackgroundService`, `Channel`, durable queue	Lost in-memory jobs
Distributed locks	Coordinate rare shared work	Lease table, TTL, fencing token	Using locks instead of constraints
Failure handling	Limit blast radius	`CancellationToken`, retries with jitter, DLQ	Retrying non-idempotent work
Cost trade-offs	Keep system economically sane	Sampling, batching, caching, capacity metrics	Complexity before evidence