Model Online / On-Site
Duration
Modules 5 modules

Course Objective & Scope

With its masterless peer-to-peer ring architecture and multi-datacenter replication, Apache Cassandra can absorb hundreds of thousands of writes per second with no single point of failure. However, designing tables like a relational schema, relying on ALLOW FILTERING or choosing a skewed Partition Key will quickly bring a cluster to its knees.

This course covers Cassandra's token ring and consistent hashing, replication strategies and tunable consistency, the LSM-Tree storage engine (CommitLog, Memtable, SSTables, Compaction), query-driven CQL data modelling, avoiding wide partitions and tombstone storms, and production cluster operations.

Who Should Attend?

  • Backend engineers building high-write messaging, IoT, time-series, ledger and activity-feed services
  • Software and data architects designing active-active, multi-datacenter storage tiers
  • DBAs, SREs and platform engineers operating Apache Cassandra (or ScyllaDB) clusters
  • Prerequisite: familiarity with core database and distributed systems concepts

Key Takeaways

  • Configure replication factors and tunable consistency (LOCAL_QUORUM) within the CAP/PACELC trade-off space
  • Model CQL tables query-first by selecting Partition Keys for even distribution and Clustering Columns for sort order
  • Tune the LSM-Tree write/read path, Bloom filters and compaction strategies (STCS, LCS, TWCS)
  • Prevent hot partitions, tombstone accumulation, secondary-index pitfalls and multi-partition scatter-gather queries
  • Configure client driver policies and run production maintenance with nodetool, repairs and snapshots

Curriculum

Cassandra Distributed Architecture and Tunable Consistency
  • Masterless peer-to-peer token ring, Consistent Hashing, VNodes and the Gossip protocol
  • SimpleStrategy vs NetworkTopologyStrategy, rack/datacenter awareness and Replication Factor (RF)
  • Tunable Consistency: ONE, QUORUM, LOCAL_QUORUM, ALL and the R + W > RF strong consistency rule
  • Hinted Handoff, Read Repair and Anti-Entropy (Merkle tree) repair mechanisms
The LSM-Tree Storage Engine: Writes, Reads and Compaction
  • The write path: CommitLog, Memtable and flushing immutable SSTables to disk
  • The read path: Memtable, Key/Row Cache, Bloom Filters, Partition Indexes and merging SSTables
  • How deletes really work: Tombstones, gc_grace_seconds and tombstone threshold errors
  • Choosing a Compaction strategy: SizeTiered (STCS), Leveled (LCS) and TimeWindow (TWCS)
Query-Driven CQL Data Modelling
  • Why there are no JOINs: designing one denormalised table per application access pattern
  • Anatomy of a Primary Key: Partition Keys (node placement) and Clustering Columns (on-disk sort order)
  • Partition sizing limits, time-bucketing to prevent unbounded wide partitions, and collection/UDT types
CQL Anti-Patterns and Client Driver Best Practices
  • Why ALLOW FILTERING, high-cardinality secondary indexes, Materialized Views and large IN queries hurt clusters
  • Lightweight Transactions (LWT / Paxos) and when BATCH statements help vs hurt
  • Token-aware, DC-aware driver policies, prepared statements and stateless cursor pagination
Cluster Operations, Monitoring and Maintenance
  • Health checks with nodetool (status, tpstats, tablestats, compactionstats) and JMX/Prometheus metrics
  • Scaling nodes in/out, scheduling anti-entropy repairs (Cassandra Reaper) and snapshot backups

Other Courses in This Category

  • Search and Real-Time Analytics with Elasticsearch and OpenSearch

    Inverted index and Lucene internals, text analysis (tokenizers/filters), mapping design, Query DSL and BM25 relevance tuning, Aggregations, hybrid/vector search and cluster lifecycle management (ILM/ISM).

    Databases Corporate
    Live Course
    5 modules
    View
  • Graph Databases, Data Modelling and Cypher with Neo4j

    Labeled Property Graph modelling, the Cypher query language and variable-length path traversals, index-free adjacency, PROFILE/EXPLAIN tuning, Graph Data Science (PageRank, Louvain) and Knowledge Graph/GraphRAG patterns.

    Databases Corporate
    Live Course
    5 modules
    View
  • Time-Series Databases with TimescaleDB and InfluxDB

    Time-series data modelling, managing high cardinality, TimescaleDB (hypertables, continuous aggregates, compression) and InfluxDB/VictoriaMetrics, downsampling and automated retention policies.

    Databases Corporate
    Live Course
    5 modules
    View
  • Vector Databases, ANN Indexing and Hybrid Search Architecture

    Dense and sparse embeddings, distance metrics, ANN index algorithms (HNSW, IVF, DiskANN), vector quantization (PQ, Scalar, Binary), metadata filtering, hybrid search and recall/latency benchmarking.

    Databases Corporate
    Live Course
    5 modules
    View
  • Practical Vector Search with pgvector, Milvus, Weaviate and Pinecone

    Hands-on engineering across the four leading vector platforms: indexing, hybrid search, multi-tenant isolation and architectural selection with PostgreSQL pgvector, Milvus, Weaviate and Pinecone.

    Databases Corporate
    Live Course
    5 modules
    View
  • Database Performance Tuning and SQL Query Optimization

    Cost-Based Optimizer internals, reading Execution Plans, SARGability, B-Tree/Covering/Filtered index design, join algorithms, parameter sniffing, lock/isolation analysis and partitioning — for developers and DBAs.

    Databases Corporate
    Live Course
    5 modules
    View

Get in touch