brianfrankcooper/ycsb

Yahoo! Cloud Serving Benchmark

View on GitHub ↗Jump to charts ↓Open shareable report

Summary Information

Updated 16 minutes ago
Added to GitGenius on September 13th, 2026
Created on April 19th, 2010
Open Issues & Pull Requests: 384 (+0)
GitHub issues: Enabled
Number of forks: 2,326
Total Stargazers: 5,231 (+0)
Total Subscribers: 208 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 32.6 hours
Mean response time: 75.5 days
90th percentile: 374.5 days
Tracked items: 19

Most active contributors

Sign in to see contributor activity.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 25
New in 7 days: 0
Closed in 7 days: 0
Avg open age: 774 days
Stale 30+ days: 25
Stale 90+ days: 25

Recent activity

Opened in 7 days: 0
Closed in 7 days: 0
Comments in 7 days: 0
Events in 7 days: 0

Top labels

No label distribution available yet.

Most active issues this week

No issue events were indexed in the last 7 days.

Detailed Description

YCSB is a benchmarking tool that measures the performance and behavior of cloud-serving databases and key-value stores under various workload patterns.

The tool addresses the need to evaluate database systems under realistic, reproducible conditions. It works by defining standardized workloads that exercise databases with configurable mixes of read, write, and scan operations, then collecting detailed performance metrics including latency percentiles. The framework includes bindings for multiple database systems, allowing the same workload to run against different backends for comparative analysis.

Teams should adopt YCSB when they need to benchmark database performance before deployment, compare different database systems objectively, or validate that a database meets latency requirements under specific access patterns. It suits projects where tail latency matters—the tool emphasizes measuring percentiles like P99, P99.9, and P99.99 rather than averages, recognizing that users experience the worst-case latencies in a distribution. The framework is particularly valuable for understanding how many clients will observe degraded performance under load, which the documentation illustrates through concrete probability calculations. Organizations can run multiple loader instances and merge results to simulate larger-scale scenarios, though the tool's documentation notes that this approach requires careful interpretation to avoid distorting the original workload distributions.

The project maintains active bindings for multiple database systems through a modular Maven-based build system, allowing contributors to add new database adapters independently. Development follows a structured approach to workload definition, with documented core properties and configuration options that enable reproducible benchmarking across different environments. The tool provides detailed guidance on latency analysis methodology, including warnings against common statistical mistakes like averaging percentiles, which reflects a mature understanding of performance measurement pitfalls.