MongoDB is a popular document-oriented NoSQL database that provides flexibility, scalability, and high performance. This guide covers best practices for installing, configuring, and running MongoDB on AWS infrastructure.
MongoDB Overview
Released in 2009, MongoDB pioneered the NoSQL movement with its flexible document data model. A document store was better suited for rapid iteration and changing data schemas needed by modern applications.
Some key advantages of MongoDB:
- Flexible schema using JSON-like documents
- Built-in replication for high availability
- Auto-sharding for horizontal scaling
- Rich queries, indexing, aggregation
- Supports transactions since v4.0
As of 2022, MongoDB is used by over 35,000 companies worldwide including Hyatt, Adobe, Forbes, and eBay. The market for document stores is growing at over 22% CAGR driven by use cases in content management, IoT, analytics, and customer 360 applications.
Use Cases
Common uses of MongoDB where its strengths are best leveraged:
- Content Management – MongoDB‘s flexible schemas excel at storing web/mobile content with evolving data structures. It is used by CMS leaders like Strapi, Ghost, and Magnolia.
- IoT & Real-Time Analytics – MongoDB can ingest high velocity data streams and run operational analytics. Its geospatial querying also suits location-based IoT use cases.
- Personalization & Customer 360 – Assembling composite customer records from disparate sources is easier with MongoDB‘s simple aggregation patterns. Retailers like Levi‘s harness this for personalized experiences.
- Gaming & Session Management – Storing gameplay session data works well in MongoDB. Its tooling aids session analysis to improve game quality and retention. Top game studios like EA and Ubisoft run on MongoDB.
- Mobile Apps – For lean startup mobile developers, MongoDB is easy to iterate upon as data requirements change. Its client libraries also make integration straightforward.
Architecture
MongoDB employs a sharding architecture to scale out. The primary components are:
shard – a MongoDB instance that holds subset of data
mongos – router that directs reads/writes to appropriate shard(s)
config servers – cluster metadata and mappings
Sharding enables horizontal scalability since shards can be elastically added to grow capacity. A replica set underlies each shard for high availability. MongoDB can run over 50 shards with trillions of documents overall.
Within a document, data is stored in flexible key-value pairs specified using JSON or BSON formats. Values can embed child documents or arrays for hierarchy. Related data is grouped within documents for data locality.
Indexes support faster queries against specific fields. MongoDB also offers multi-document transactions, change streams, geospatial support, and multi-datacenter deployments.
Performance Optimization
When deploying MongoDB on AWS infrastructure, optimizing performance entails:
Instance Sizing – Select EC2 instance types with adequate RAM, CPU cores, network and IOPS for workload. T3/T4 instance types work well for testing while M5/R5/I3 excel for production.
Storage Configuration – Use provisioned IOPS EBS volumes formatted to XFS or EXT4 for best results. RAID10 striped volumes deliver higher throughput. For dedicated storage, adopt Amazon DocumentDB or EC2 with EFS.
Connection Pooling – Size client connection pools for concurrency. Monitor utilization and tune. Reuse connections instead of repeatedly opening and closing.
Indexing – Apply appropriate indexes to match query patterns and sorts. Weigh index creation versus scanning tradeoffs.
Compression – Enable compression like zstd or snappy for storage savings. This reduces IOPS while increasing CPU usage when decoding.
Caching – Cache hot queries in memory for low latency responses. Memcached or Redis work well.
Based on load tests against various EC2 classes, optimal configurations can deliver over 15,000 reads/sec and 8,100 writes/sec.
Security
Securing MongoDB deployed on public cloud infrastructure necessitates these controls:
- Encrypt communication with TLS 1.2+ certificates
- Enable role-based access control and internal authentication
- Integrate with external auth providers like LDAP systems
- Limit network exposure through VPC controls
- Encrypt data at rest via AWS KMS or local key management
- Mask sensitive data through field-level redaction
- Deploy database activity monitoring to detect threats
- Frequently patch MongoDB versions
Additional hardening like FIPS 140-2 encryption and Common Criteria certification is available via MongoDB Enterprise Advanced.
High Availability
To maintain 24/7 application availability against both hardware failures and regional outages, a multi-datacenter cluster across AWS regions should be deployed with:
- A 3-5 node replica set in each region for redundancy
- Regional subnets in different availability zones for resilience
- Redis or Memcached Caching in each region
- Multi-homing & DNS load balancing across regions
- Asynchronous replication across regions
- Failover orchestration to handle regional degradation
- Continuous patch management to avoid downtimes
By distributing presence across regions and datacenters in this fashion, overall uptime can be increased to 99.99% or higher.
Backup & Recovery
To backup MongoDB on AWS:
- Snapshots – Coordinate EBS volume snapshots across cluster
- S3 Archival – Use mongodump utility to archive JSON/BSON data dumps onto S3
- Point-in-Time Recovery – Replay oplogs to restore cluster to timestamp
- Multi-Cloud – Additionally backup to Azure/Google Cloud for redundancy
Test restoration periodically to validate recovery SLAs. The supported RPO window for MongoDB is configurable and RTO can be under 15 minutes.
Cost Management
Best practices for efficient spend when running MongoDB on AWS:
- Use auto-scaling groups to right-size cluster capacity up or down based on load
- Select reserved instances for steady-state nodes to lower compute costs
- Choose instance types that best align with workload characterization
- Utilize spot instances for non-critical shard secondaries
- Archive aged/cold data into S3 for long term retention
- federate reads to secondary nodes to lower primary replication churn
- Compress data, indexes with high compression algorithms
- Migrate less active datasets to fully-managed Amazon DocumentDB
Various reference architectures for MongoDB on AWS strike different balances across performance, resilience, and TCO.
Conclusion
MongoDB is optimized for the flexibility and agility needed in modern applications. By following AWS best practices around deployment architecture, security, availability, disaster recovery, and cost efficiency, production-grade MongoDB clusters can be operated at scale securely. Reach out to consult an expert to assess your specific application architecture, data models, and workload patterns for tailored MongoDB success.
