Aerospike Hybrid Memory Architecture: Indexes in DRAM, Records on NVMe
How Aerospike’s HMA stores indexes in DRAM and records on NVMe delivers sub-millisecond latency at lower cost, with concrete sizing and hardware guidance.
Loading…
ReadMeFeed / Community knowledge
#storage-engine
Real questions. Useful conversations. Find the people who know your stack.
aerospikeChoosing Between InnoDB and Aria in MariaDB
Decide whether to use InnoDB or Aria for a MariaDB table based on transactions, foreign keys, locking, and memory.
Aerospike memory-size vs data-size: Choosing a Namespace Storage Engine
Choosing between Aerospike's memory and device namespaces is a capacity-versus-latency decision with a hidden RAM cost: the primary index always lives in memory.
Choosing RocksDB Compaction Style for Write-Heavy Workloads: Level vs Universal
A decision guide for selecting RocksDB compaction style under high ingest rates and limited SSD endurance, with configuration and validation steps.
Isolating Workloads in RocksDB with Column Families
Learn how to partition RocksDB data into Column Families, tune MemTable and compaction settings independently, and avoid common initialization and resource-leak pitfalls.
Architecture Note: Using RocksDB Leveled Compaction for Write‑Heavy Workloads
An architecture note outlining the minimal RocksDB configuration for write‑heavy workloads, covering requirements, design, trust boundaries, operational checks, failure modes, and triggers for redesign.
Using RocksDB Column Families to Isolate Workloads in a Single DB Instance
Learn how to create, configure, and use multiple RocksDB column families to isolate workloads while sharing a single database instance.
cppOrganizing RocksDB Data: When to Use Column Families Over Key Prefixes
Stop treating RocksDB as a single bucket. Learn how Column Families provide isolated namespaces, independent compaction, and atomic writes to optimize your storage layer.
Choosing a Compression Algorithm in RocksDB: A Decision Guide
A decision guide for selecting RocksDB compression algorithms (Snappy, LZ4, Zstd, etc.) with per‑column‑family tuning, a concrete C++ config example, and a reproducible db_bench micro‑benchmark to validate trade‑offs.