Tag
GreptimeDB explains its architecture of separating compute and storage via shared object storage in distributed deployments, and describes how it reuses SST data files during repartitioning to avoid migrating historical data wholesale, while also comparing the differences with ClickHouse.
EMQX reports on a benchmark showing GreptimeDB v1.2.0 achieving 31% higher ingestion throughput than InfluxDB 3.11.1 Enterprise, along with query and storage comparisons relevant to IoT telemetry on EMQX Tables.
GreptimeDB v1.3 introduces experimental provider-neutral AI functions in SQL — ai_match, ai_choose, and ai_score — allowing users to filter, classify, and rate text rows using natural-language prompts via TypeSafe's Jev model backend.
GreptimeDB v1.1 introduces online repartitioning for existing tables via a single ALTER TABLE statement, eliminating the need for data migration, dual-writes, or application changes. It leverages shared object storage and logical shards to update manifests and routing without moving data between nodes.
Xiaomi's smart factory replaced Loki with GreptimeDB for log storage, handling billions of rows per month with tailored indexing: Bloom skipping for high-cardinality trace_id, inverted index for low-cardinality fields, and fulltext search on message bodies.
GreptimeDB's Mito2 storage engine uses an LSM-tree design with columnar Parquet SSTs, three-level scan pruning, and TWCS compaction. The blog post provides a full walkthrough of its architecture.
GreptimeDB v1.0 introduces Pending Rows Batcher, a three-stage pipeline that moves CPU-intensive work off the Datanode's critical section, improving Prometheus remote write throughput from 1.20M to 2.17M points/sec and reducing Datanode CPU usage by 20%.