Redis: Faster RDB Loads and Stream Insight
Two merged optimizations sharply cut RDB load time via faster decompression and checksumming, while a new INFO streams section adds low-cost visibility into consumer-group backlogs.
Duration: PT2M14S
Episode overview
This episode is a short developer briefing from Redis.
It explains recent repository work in plain language.
- Show: Redis
- Published: 2026-10-15T13:29:03Z
- Audio duration: PT2M14S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Hello, today is October 15th, 2026, and this is your Redis developer briefing.
The big pattern is faster restarts. Two merged changes attack RDB loading from opposite ends — decompression and checksumming — cutting load times by roughly fourteen to twenty-six percent in reported benchmarks.
First, PR 15765 replaces the branch-heavy L Z F decoder for RDB loads with a leaner, L Z four style decoder that copies in fixed-size chunks. The old decoder remains as a test oracle for a new randomized fuzz test. Isolated decode is reported at two to three times faster, with full RDB loads sixteen to forty-nine…
Second, PR 15732 accelerates C R C 64 checksumming with hardware carry-less multiplication, about six times faster for larger buffers, plus batched checksums and direct reads for large values to avoid extra copies. Small-key datasets improved about fourteen percent, large-value datasets about twenty-six percent. The…
The second theme is stream observability. PR 15682 adds a new INFO streams section with base-two histograms for groups per stream, pending list size per group, and consumers per group. It is off by default behind a modifiable stream stats flag, maintained incrementally to…
Fin…
Nearby episodes from Redis
- Float Formatting Speedup and Restart Fix
- Replication Consistency and macOS Build Repair
- Native Roaring Bitmaps Land
- Replica Write Loss and Stream Fixes
- Probabilistic Types Move Into Core
- Silent Replica Drift and Persistence Fixes
- Native Probabilistic Types and Hot-Path Trimming
- Migration Progress Visibility and Core Operation Fixes