Skip to main content

Questions

  • what is opslog in mongodb, location of opslog in mongodb
  • what is mongodump and if we take backup through mongodump so it will take backup of all the databases or particular databases
  • how do you create replicaset performance matrix in mongodb
  • how to check replicaset is sync or not
  • user roles how do you set roles our database is not connected with outside network security authorisation configuration file
  • difference between sharding and replicaset
  • how to configure replica set
  • capped collections
  • How do you optimize a MongoDB query performance?
  • How would you handle a situation where a replica set member is consistently lagging behind the others?
  • How do you enable authentication in MongoDB?
  • How can you create an index in MongoDB, If we have 100 of mongodb servers?
  • How to do monitoring of all the databases the default interval to write updates to the disk?
  • RAM usage in mongodb db.stat() Monitoring tools and features
  • Mongodb TTL index
  • Mongodb Change streams
  • Mongodb read and write concern
  • Mongodb profiler - Database Profiler - MongoDB Manual
  • Mongodb Pipeline

Top 10 MongoDB Questions for a Senior Database Architect​

1. Schema Design & Modeling​

Q: MongoDB is schema-less, but in production we still need schema planning. How would you design collections for a high-write, high-read analytics platform that needs both transactional consistency and flexible query capabilities? What trade-offs would you consider between embedding vs. referencing?

A:

  • Approach:
    • Use schema versioning to track document evolution over time.
    • For high-write workloads → Avoid deep nesting; keep documents under 16MB BSON limit.
    • Embedding: Good for one-to-few relationships, avoids joins, improves read performance.
    • Referencing: Better for many-to-many relationships, reduces document growth and avoids hot document locking.
  • Example Decision:
    • User profiles: embed address & preferences (read together often).
    • Transactions: separate collection with reference to user_id (to allow indexing & partitioning).
  • Trade-offs:
    • Embedding = faster reads but potential for large doc updates.
    • Referencing = smaller docs, more flexible but requires $lookup joins.
  • Performance note: Pre-compute and denormalize only where query speed is critical.

2. Sharding Strategy​

Q: How would you choose a shard key for a multi-terabyte, globally distributed MongoDB cluster?

A:

  • Principles:
    • High cardinality (to spread data evenly).
    • Avoid monotonically increasing values (like timestamps) to prevent hotspot shards.
    • Consider zone sharding for geo-distribution.
  • Process:
    • Analyze current query patterns & write distribution (system.profile, collStats).
    • Test candidate keys in staging with sh.status() and chunk distribution metrics.
  • Example:
    • E-commerce orders: (region_code, order_id) compound key — region controls geo-locality, order_id ensures randomness.
  • Pitfall: Avoid sharding on fields that change frequently — causes expensive document migrations.

3. Indexing Optimization​

Q: How to identify & resolve performance bottlenecks from inefficient indexes?

A:

  • Use db.collection.explain("executionStats") to check IXSCAN vs COLLSCAN.
  • Remove unused indexes (db.collection.getIndexes() + MongoDB Atlas Index suggestions).
  • Compound indexes: Match query prefix fields order to improve coverage.
  • Partial indexes: For filtered queries (e.g., only status:"active" docs).
  • Wildcard indexes: For dynamic field names in semi-structured data.
  • Pitfall: Too many indexes slow writes — balance read vs write performance.

4. Replication & High Availability​

Q: How to ensure minimal downtime in a mission-critical financial app?

A:

  • Use Replica Set with 5+ members (e.g., 3 voting, 2 hidden backups).
  • Write Concern: "majority" to ensure durability.
  • Read Preference: "nearest" or "secondaryPreferred" for global reads.
  • Use hidden members for analytics to offload queries.
  • Deploy arbiter only if required for voting quorum (avoid if possible).
  • Pitfall: Never run majority writes with low wtimeout without understanding rollback implications.

5. Aggregation Pipeline Performance​

Q: How to optimize aggregation pipelines for billions of docs?

A:

  • Always $match & $project early to reduce document size.
  • Use covered queries when possible.
  • For $lookup, limit joined dataset with $match on both sides.
  • Avoid unbounded $group — use $bucket or $bucketAuto for summarization.
  • For memory-intensive stages, ensure indexes align with $sort fields to avoid allowDiskUse overhead.
  • Consider pre-aggregated collections via $merge for hot dashboards.

6. Transactions in MongoDB​

Q: When not to use multi-document ACID transactions?

A:

  • When:
    • Writes can be modeled as single-document atomic updates.
    • For high-throughput writes — transactions add ~30–50% latency.
    • Large multi-collection updates in high-traffic systems.
  • Alternatives:
    • Use two-phase commit pattern for critical updates.
    • Restructure schema for single-document updates.

7. Data Lifecycle Management​

Q: How to implement time-series retention in IoT telemetry?

A:

  • Use Time Series Collections (createCollection with timeseries option).
  • Configure TTL Index on the timestamp field: db.telemetry.createIndex({ ts: 1 }, { expireAfterSeconds: 7776000 }); // 90 days
  • For high-ingest IoT:
    • Partition by time & device_id.
    • Archive expired data to S3 via Change Streams.

8. Security & Compliance​

Q: How to make MongoDB HIPAA-compliant?

A:

  • RBAC: Create roles per microservice with least privilege.
  • Encryption: Enable Encryption At Rest (KMIP or AWS KMS) + TLS for in-transit.
  • Field-Level Encryption: Encrypt sensitive fields (SSN, PHI) with client-side keys.
  • Auditing: Enable auditLog to track access.
  • Test performance hit — field-level encryption can double document size.

9. Operational Monitoring & Troubleshooting​

Q: How to monitor real-time MongoDB performance?

A:

  • Tools: mongostat, mongotop, Atlas metrics dashboard, db.currentOp().

  • Key metrics:

    • Replication lag
    • Opcounters (reads/writes per sec)
    • Cache hit ratio (wiredTiger.cache)
    • Slow queries from profiler.

  • Actions:

    • Optimize slow queries first.
    • Rebalance chunks if uneven.
    • Increase WiredTiger cache if hitting read misses.

10. Migration & Upgrade Strategies​

Q: How to upgrade a 10TB v4.2 cluster to v6.0 with minimal downtime?

A:

  • Prep:

  • Plan:

    • Rolling upgrade on replica set (secondary → primary).
    • Monitor replication lag before stepping down primary.

  • Validation:

    • Run dbHash before & after upgrade.
    • Check app logs & profiler for query plan changes.

  • Pitfall: Do not skip major versions — must upgrade sequentially (4.2 → 4.4 → 5.0 → 6.0).