#apache-cassandra

Building Menu Vision: Real-Time Dish Recognition

Building Menu Vision: Real-Time Dish Recognition Arpitha Dudi, Growth Tech Lead Aug 20, 2026 Menus aren’t just lists, they’re a window into a restaurant’s unique offerings, specialties, and personality, shaping...

The road to ACID transactions in Cassandra 6

Cassandra 6 đang phát triển tính năng giao dịch ACID, nâng cao khả năng xử lý nhất quán dữ liệu trong hệ thống phân tán.

Lập trình viên cần đọc bài này để hiểu cách Cassandra 6 mở rộng khả năng quản lý dữ liệu đồng thời (ACID) cho hệ thống phân tán, giúp tối ưu hóa sự tin cậy và nhất quán trong ứng dụng khi sử dụng cơ sở dữ liệu phân tán.

Advanced data modeling: Using user-defined types and Protocol Buffers for Amazon Keyspaces

Amazon Keyspaces supports two approaches for modeling complex, nested data: user-defined types (UDTs) and Protocol Buffers (Protobuf). UDTs are native CQL structures that enable server-side schema validation and field-level querying, but cannot be altered after creation and have nesting limits. Protobuf stores data as binary BLOBs, enabling schema evolution, cross-language portability, and compact storage, but lacks server-side validation. The post walks through creating UDTs with nested collections, querying and updating them, using lightweight transactions, and implementing Protobuf serialization in Java with Amazon Keyspaces BLOB columns. A real estate property management system is used as the example domain throughout.

Migrate to Amazon Keyspaces with zero downtime using ZDM Proxy

A detailed guide on migrating from self-managed Apache Cassandra to Amazon Keyspaces with zero downtime using the ZDM (Zero Downtime Migration) Proxy. The proxy intercepts CQL traffic at the network layer, fanning out writes to both the origin cluster and Amazon Keyspaces simultaneously, requiring no application code changes. The solution deploys the proxy on AWS Fargate behind a Network Load Balancer using a single CloudFormation stack. The migration follows six phases: preparation, historical data loading with CQLReplicator, enabling dual writes, validation (dual reads with discrepancy monitoring), cutover (blue-green or canary), and decommission. Security considerations include TLS via socat, credentials in AWS Secrets Manager, least-privilege IAM roles, and VPC endpoints. Cost components during migration include Fargate tasks (~$100/month for 3 tasks), NLB ($20–50/month), and AWS Glue for CQLReplicator.

Extremely fast LSM Trees

Extremely fast LSM Trees In my latest piece of writing about Write Buffering, I mentioned LSM Trees in one paragraph. Let’s unpack what exactly LSM trees are, what purpose do they serve, and why …

Netflix Cuts Cassandra Read Latency from Seconds to Milliseconds with Dynamic Partition Splitting

Netflix engineers built a dynamic partition-splitting mechanism for Apache Cassandra to tackle wide partitions in time-series workloads. The system automatically detects oversized partitions exceeding predefined thresholds and asynchronously splits them into smaller child partitions, maintaining a metadata layer to track parent-child relationships and route reads transparently. Results from multiple child partitions are merged before returning to clients, requiring no application changes or downtime. After deployment on Netflix's TimeSeries Abstraction platform, average read latency dropped from seconds to low double-digit milliseconds, tail latencies fell to under 200ms, and CPU utilization and read timeouts decreased significantly. The initial implementation focused on immutable partitions for safety, with phased rollouts and validation pipelines comparing old and new read paths. Future work includes support for mutable partitions and reprocessing of failed splits.