feat: Update performance metrics to 2.42M RPS with improved latency

Ignacio Van Droogenbroeck · claude · Ignacio Van Droogenbroeck · commit 2390f90f1576 · 2025-10-15T19:08:58.000-03:00
Update throughput from 2.01M/2.32M to 2.42M records/sec across all
documentation to reflect latest benchmarks with authentication enabled
and token caching.

Updated latency metrics:
- p50: 6.75ms → 1.74ms (78x improvement vs row format)
- p95: 39.46ms → 28.13ms (30x improvement vs row format)
- p99: 59.09ms → 45.27ms (34x improvement vs row format)

Updated files:
- Homepage and features components
- Introduction and getting started guides
- Performance benchmarks documentation
- Installation guides (Docker and native)
- Advanced features (WAL, compaction)
- Integration guides

Co-Authored-By: Claude &lt;noreply@anthropic.com&gt;
diff --git a/README.md b/README.md
@@ -4,7 +4,7 @@ Official documentation for [Arc](https://github.com/basekick-labs/arc) - the fas
 
 ## About
 
-This repository contains the documentation website for Arc, a high-performance time-series data warehouse built on DuckDB and Parquet. Arc achieves 36.43s on ClickBench (99.9M rows) and delivers 2.01M records/sec ingestion.
+This repository contains the documentation website for Arc, a high-performance time-series data warehouse built on DuckDB and Parquet. Arc achieves 36.43s on ClickBench (99.9M rows) and delivers 2.42M records/sec ingestion.
 
 Visit the live documentation at [docs.basekick.net](https://docs.basekick.net)
 
diff --git a/docs/advanced/compaction.md b/docs/advanced/compaction.md
@@ -25,10 +25,10 @@ Compaction is **enabled by default** and runs automatically every hour.
 
 ### The Small File Problem
 
-Arc's high-performance ingestion (2.01M records/sec) creates many small files:
+Arc's high-performance ingestion (2.42M records/sec) creates many small files:
 
 ```
-At 2.01M records/sec with 5-second flush:
+At 2.42M records/sec with 5-second flush:
 → 10M records every 5 seconds
 → 12 files per minute per measurement
 → 720 files per hour per measurement
diff --git a/docs/advanced/wal.md b/docs/advanced/wal.md
@@ -21,7 +21,7 @@ There are **two different WAL features** in Arc:
 WAL is an optional durability feature that persists all incoming data to disk **before** acknowledging writes. When enabled, Arc guarantees that data can be recovered even if the instance crashes.
 
 :::info
-WAL is **disabled by default** to maximize throughput (2.01M records/sec). Enable it when zero data loss is required.
+WAL is **disabled by default** to maximize throughput (2.42M records/sec). Enable it when zero data loss is required.
 :::
 
 ### When to Enable WAL
@@ -32,15 +32,15 @@ Enable WAL if you need:
 - **Recovery from unexpected failures** (power loss, OOM kills)
 
 Keep WAL disabled if you:
-- **Prioritize maximum throughput** (2.01M records/sec)
+- **Prioritize maximum throughput** (2.42M records/sec)
 - **Can tolerate 0-5 seconds data loss** on rare crashes
 - **Have client-side retry logic** or message queue upstream
 
 ### Performance vs Durability Tradeoff
 
 | Configuration | Throughput | Data Loss Risk |
 |--------------|-----------|----------------|
-| **No WAL (default)** | 2.01M rec/s | 0-5 seconds |
+| **No WAL (default)** | 2.42M rec/s | 0-5 seconds |
 | **WAL + async** | 1.67M rec/s (-17%) | &lt;1 second |
 | **WAL + fdatasync** | 1.63M rec/s (-19%) | Near-zero |
 | **WAL + fsync** | 1.67M rec/s (-17%) | Zero |
@@ -403,7 +403,7 @@ $ du -sh ./data/wal
 ### Performance Degradation with WAL
 
 **Symptoms:**
-- Throughput dropped from 2.01M to 600K rec/s
+- Throughput dropped from 2.42M to 600K rec/s
 - High CPU usage from fsync calls
 
 **Solutions:**
diff --git a/docs/installation/docker.md b/docs/installation/docker.md
@@ -171,7 +171,7 @@ docker-compose up -d --build
 
 ## Performance Considerations
 
-Docker deployment achieves **~570K records/sec** write throughput. For maximum performance (2.01M RPS), use [native deployment](/arc/installation/native).
+Docker deployment achieves **~570K records/sec** write throughput. For maximum performance (2.42M RPS), use [native deployment](/arc/installation/native).
 
 ### Optimize Docker Performance
 
diff --git a/docs/installation/native.md b/docs/installation/native.md
@@ -4,7 +4,7 @@ sidebar_position: 2
 
 # Native Installation
 
-Install Arc natively for **maximum performance**. Native deployment achieves **2.01M records/sec** (3.5x faster than Docker).
+Install Arc natively for **maximum performance**. Native deployment achieves **2.42M records/sec** (3.5x faster than Docker).
 
 ## Prerequisites
 
@@ -203,8 +203,8 @@ pip install uvloop httptools orjson
 | Backend | Performance | Use Case |
 |---------|-------------|----------|
 | **Local NVMe** | 2.08M RPS | Maximum performance, single-node |
-| **Local SSD** | 2.01M RPS | High performance, single-node |
-| **MinIO** | 2.01M RPS | Distributed, multi-tenant |
+| **Local SSD** | 2.42M RPS | High performance, single-node |
+| **MinIO** | 2.42M RPS | Distributed, multi-tenant |
 | **AWS S3** | 1.5M RPS | Cloud, unlimited scale |
 
 **Recommendation**:
diff --git a/docs/integrations/telegraf.md b/docs/integrations/telegraf.md
@@ -13,7 +13,7 @@ Arc is compatible with InfluxDB Line Protocol, making it a perfect drop-in repla
 **Benefits:**
 - No configuration changes needed
 - All Telegraf plugins work unchanged
-- 8x faster than InfluxDB (2.01M RPS)
+- 8x faster than InfluxDB (2.42M RPS)
 - Lower cost with object storage
 - Full SQL analytics with DuckDB
 
diff --git a/docs/intro.md b/docs/intro.md
@@ -5,11 +5,11 @@ slug: /
 
 # Welcome to Arc
 
-Arc is a high-performance time-series data warehouse built on DuckDB, Parquet, and flexible storage backends. It achieves **2.01M records/sec ingestion** and is the **fastest time-series database**.
+Arc is a high-performance time-series data warehouse built on DuckDB, Parquet, and flexible storage backends. It achieves **2.42M records/sec ingestion** and is the **fastest time-series database**.
 
 ## Key Features
 
-- **High-Performance Ingestion**: 2.01M records/sec with MessagePack binary protocol
+- **High-Performance Ingestion**: 2.42M records/sec with MessagePack binary protocol
 - **Fast Analytical Queries**: Powered by DuckDB with full SQL support
 - **Flexible Storage**: Local filesystem, MinIO, AWS S3, or Google Cloud Storage
 - **Multi-Database Architecture**: Organize data by environment, tenant, or application
diff --git a/docs/performance/benchmarks.md b/docs/performance/benchmarks.md
@@ -193,16 +193,16 @@ Arc achieves exceptional write throughput through MessagePack columnar binary pr
 
 | Wire Format | Throughput | p50 Latency | p95 Latency | p99 Latency | Notes |
 |-------------|------------|-------------|-------------|-------------|-------|
-| **MessagePack Columnar** | **2.32M RPS** | **6.75ms** | **39.46ms** | **59.09ms** | Zero-copy passthrough (RECOMMENDED) |
+| **MessagePack Columnar** | **2.42M RPS** | **1.74ms** | **28.13ms** | **45.27ms** | Zero-copy passthrough + auth cache (RECOMMENDED) |
 | **MessagePack Row** | **908K RPS** | **136.86ms** | **851.71ms** | **1542ms** | Legacy format with conversion overhead |
 | **Line Protocol** | **240K RPS** | N/A | N/A | N/A | InfluxDB compatibility mode |
 
 **Columnar Format Advantages:**
-- **2.55x faster throughput** vs row format (2.32M vs 908K RPS)
-- **20x lower p50 latency** (6.75ms vs 136.86ms)
-- **21x lower p95 latency** (39.46ms vs 851.71ms)
-- **26x lower p99 latency** (59.09ms vs 1542ms)
-- **67x fewer errors** under load (63 vs 4,211 errors at 2.5M target RPS)
+- **2.66x faster throughput** vs row format (2.42M vs 908K RPS)
+- **78x lower p50 latency** (1.74ms vs 136.86ms)
+- **30x lower p95 latency** (28.13ms vs 851.71ms)
+- **34x lower p99 latency** (45.27ms vs 1542ms)
+- **Near-zero authentication overhead** with 30s token cache
 
 **Test Configuration**:
 - Hardware: Apple M3 Max (14 cores)
diff --git a/src/components/HomepageFeatures/index.tsx b/src/components/HomepageFeatures/index.tsx
@@ -22,10 +22,10 @@ const FeatureList: FeatureItem[] = [
   },
   {
     title: 'Massive Throughput',
-    icon: '2.01M',
+    icon: '2.42M',
     description: (
       <>
-        Ingest 2.01M records/sec with MessagePack binary protocol.
+        Ingest 2.42M records/sec with MessagePack binary protocol.
         Built for IoT sensors, metrics collection, and observability platforms that need extreme write performance.
       </>
     ),
diff --git a/src/pages/index.tsx b/src/pages/index.tsx
@@ -31,7 +31,7 @@ function HomepageHeader() {
           </Link>
         </div>
         <div style={{marginTop: '2rem', fontSize: '1.2rem', opacity: 0.9}}>
-          <strong>36.43s</strong> ClickBench cold run • <strong>2.01M</strong> records/sec • <strong>205+</strong> GitHub stars
+          <strong>36.43s</strong> ClickBench cold run • <strong>2.42M</strong> records/sec • <strong>205+</strong> GitHub stars
         </div>
       </div>
     </header>
@@ -42,7 +42,7 @@ export default function Home(): ReactNode {
   return (
     <Layout
       title="Arc - Fastest Time-Series Database"
-      description="Arc is the fastest time-series database. 36.43s cold run on ClickBench, 2.01M records/sec ingestion, powered by DuckDB and Parquet.">
+      description="Arc is the fastest time-series database. 36.43s cold run on ClickBench, 2.42M records/sec ingestion, powered by DuckDB and Parquet.">
       <HomepageHeader />
       <main>
         <HomepageFeatures />
