Repository navigation
Expand file tree
/
Copy pathindex.html
More file actions
364 lines (339 loc) · 20.7 KB
/
Copy pathindex.html
File metadata and controls
364 lines (339 loc) · 20.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
---
layout: default
title: "Multikernel Technologies - Split-Kernel Architecture for Linux"
description: "Multikernel turns servers you own and VMs you rent into a private cloud: bare-metal performance, kernel-level isolation, no hypervisor, no virtualization tax."
i18n: true
---
<main>
<!-- Hero Section -->
<section class="hero">
<div class="hero-container">
<div class="hero-badge">
<i data-lucide="cpu"></i>
<span>Split-Kernel Architecture</span>
</div>
<h1>Your Cloud. Your Hardware. No Virtualization Tax.</h1>
<p class="subtitle">Build a private cloud on servers you own, or bring your own cloud (BYOC) and reclaim the VMs you already rent. Every workload runs in its own Linux kernel with dedicated cores: bare-metal performance, kernel-level isolation, no hypervisor, no shared kernel attack surface.</p>
<div class="button-group">
<a href="https://github.com/multikernel" target="_blank" rel="noopener noreferrer" class="btn btn-primary">
<i data-lucide="github"></i>
View on GitHub
</a>
<a href="https://calendar.app.google/nc1upkqQoUeoukdD9" target="_blank" rel="noopener noreferrer" class="btn btn-secondary">
<i data-lucide="calendar"></i>
Schedule a Demo
</a>
</div>
<div class="hero-stats">
<a class="hero-stat" href="/2026/08/16/multikernel-vs-kvm-lmbench/">
<span class="hero-stat-number">2.5x</span>
<span class="hero-stat-label">Faster context switches than KVM</span>
<span class="hero-stat-detail">1.37 µs vs 3.42 µs</span>
</a>
<a class="hero-stat" href="/2026/09/06/multikernel-vs-kvm-iperf/">
<span class="hero-stat-number">10x</span>
<span class="hero-stat-label">Less CPU to receive at line rate</span>
<span class="hero-stat-detail">vs KVM with vhost-net</span>
</a>
<a class="hero-stat" href="/2026/08/25/memory-pressure-tail-latency/">
<span class="hero-stat-number">153x</span>
<span class="hero-stat-label">Lower p99 beside a noisy neighbor</span>
<span class="hero-stat-detail">232 ms vs 35.6 s</span>
</a>
</div>
<div class="hero-trust">
<p class="trust-text">Built on upstream Linux, 100% open source</p>
<div class="trust-features">
<div class="trust-feature">
<i data-lucide="server"></i>
<span>Your Hardware or Your Cloud</span>
</div>
<div class="trust-feature">
<i data-lucide="box"></i>
<span>Docker Compatible</span>
</div>
<div class="trust-feature">
<i data-lucide="refresh-cw"></i>
<span>Zero-Downtime Updates</span>
</div>
<div class="trust-feature">
<i data-lucide="github"></i>
<span>Open Source</span>
</div>
</div>
</div>
</div>
</section>
<!-- Why Split-Kernel -->
<section id="value-proposition" class="value-section">
<div class="section-container">
<div class="value-content">
<div class="value-text">
<div class="section-badge">
<i data-lucide="layers"></i>
<span>Why Split-Kernel</span>
</div>
<h2 class="section-title">Isolation Built into the Architecture</h2>
<p class="section-subtitle">In a traditional Linux kernel, application code and device processing compete for the same CPU cores. The split-kernel architecture removes this contention by offloading all device processing to a dedicated kernel on dedicated cores. Applications see zero device interrupts, not because of tuning, but because device interrupts physically cannot reach them.</p>
<div class="value-cta">
<a href="/technology.html" class="btn-link">
Learn more about the architecture
<i data-lucide="arrow-right"></i>
</a>
</div>
</div>
<div class="value-visual">
<div class="split-diagram">
<div class="diagram-side diagram-traditional">
<div class="diagram-label">Traditional</div>
<div class="diagram-box diagram-box-shared">
<span class="diagram-tag">Single Kernel</span>
<div class="diagram-items">
<span class="diagram-item item-app">App</span>
<span class="diagram-item item-driver">Drivers</span>
<span class="diagram-item item-driver">IRQs</span>
</div>
<div class="diagram-caption">All compete for the same cores</div>
</div>
</div>
<div class="diagram-arrow">
<i data-lucide="arrow-right"></i>
</div>
<div class="diagram-side diagram-multikernel">
<div class="diagram-label">Split-Kernel</div>
<div class="diagram-box diagram-box-app">
<span class="diagram-tag">App-Kernel</span>
<div class="diagram-items">
<span class="diagram-item item-app">App</span>
<span class="diagram-item item-app">App</span>
<span class="diagram-item item-app">App</span>
</div>
<div class="diagram-caption">Zero device interrupts</div>
</div>
<div class="diagram-box diagram-box-device">
<span class="diagram-tag">Device-Kernel</span>
<div class="diagram-items">
<span class="diagram-item item-driver">Drivers</span>
<span class="diagram-item item-driver">IRQs</span>
<span class="diagram-item item-driver">I/O</span>
</div>
<div class="diagram-caption">Dedicated cores</div>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<!-- Measured Results -->
<section id="performance" class="proof-section">
<div class="section-container">
<div class="section-header">
<div class="section-badge">
<i data-lucide="gauge"></i>
<span>Benchmarks</span>
</div>
<h2 class="section-title">Measured on the Same Machine</h2>
<p class="section-subtitle">Same server, same kernel build, one variable at a time: a multikernel instance against a fully accelerated KVM guest, or two kernels against one. Every number links to its setup, caveats, and reproduction steps.</p>
</div>
<div class="proof-grid">
<a class="proof-card" href="/2026/08/16/multikernel-vs-kvm-lmbench/">
<span class="proof-number">2.5x</span>
<span class="proof-label">Faster context switches than KVM</span>
<span class="proof-compare"><strong>1.37 µs</strong> vs 3.42 µs</span>
<p class="proof-detail">Against a guest with EPT, APICv, and pinned vCPUs. KVM closes the gap only by spinning its idle cores, at up to 19 W per guest. Multikernel stays fast while sleeping in real C-states.</p>
<span class="proof-source">lmbench vs KVM <i data-lucide="arrow-right"></i></span>
</a>
<a class="proof-card" href="/2026/09/06/multikernel-vs-kvm-iperf/">
<span class="proof-number">10x</span>
<span class="proof-label">Less CPU to receive at line rate</span>
<span class="proof-compare"><strong>21%</strong> vs 215% of a core</span>
<p class="proof-detail">Both sides hit line rate. The VM paid 2x the CPU to send, 10x to receive, and added four times as much round-trip latency (0.40 ms vs 0.10 ms) through vhost-net.</p>
<span class="proof-source">iperf vs KVM <i data-lucide="arrow-right"></i></span>
</a>
<a class="proof-card" href="/2026/08/17/multikernel-will-it-scale/">
<span class="proof-number">2.6x</span>
<span class="proof-label">More throughput from the same 24 cores</span>
<span class="proof-compare"><strong>780K</strong> vs 300K unlinks/s</span>
<p class="proof-detail">Two kernels instead of one, on identical hardware: 2.1x on stat() of a shared file, 4.1x on unlink across sockets. Locks that are not shared are not contended. Control benchmarks moved by at most 1%.</p>
<span class="proof-source">will-it-scale <i data-lucide="arrow-right"></i></span>
</a>
<a class="proof-card" href="/2026/08/25/memory-pressure-tail-latency/">
<span class="proof-number">153x</span>
<span class="proof-label">Lower p99 beside a noisy neighbor</span>
<span class="proof-compare"><strong>232 ms</strong> vs 35.6 s</span>
<p class="proof-detail">A latency-sensitive task next to a memory-hungry batch job. On one kernel, cgroups protect the victim only by starving the batch job to 10 passes. With two kernels, both run at full speed.</p>
<span class="proof-source">Tail latency <i data-lucide="arrow-right"></i></span>
</a>
<a class="proof-card" href="/2026/10/14/before-the-kernel-boots-memory-setup/">
<span class="proof-number">61x</span>
<span class="proof-label">Faster large-memory instance setup</span>
<span class="proof-compare"><strong>111 ms</strong> vs 6.81 s, 32 GiB preallocated</span>
<p class="proof-detail">QEMU with preallocated RAM spends seconds populating pages before a vCPU runs, and pinning memory for a passthrough device costs the same. Multikernel assigns physical ranges it already owns, in about 110 ms from 1 GiB to 32 GiB.</p>
<span class="proof-source">Instance memory setup <i data-lucide="arrow-right"></i></span>
</a>
<a class="proof-card" href="/2026/09/30/the-kernel-crashed-the-memory-didnt/">
<span class="proof-number">0</span>
<span class="proof-label">Bytes lost across four kernel panics</span>
<span class="proof-compare"><strong>4 of 4</strong> panics verified intact</span>
<p class="proof-detail">An app-kernel panicked mid-write four times. Each time, a fresh kernel booted on the same DAXFS memory and found every byte in place, down to a half-written page. No log replay, no reload.</p>
<span class="proof-source">Crash recovery <i data-lucide="arrow-right"></i></span>
</a>
</div>
<p class="proof-footnote"><a href="/performance.html">See all benchmark results</a></p>
</div>
</section>
<!-- Who It's For -->
<section id="use-cases" class="how-it-works">
<div class="section-container">
<div class="section-header">
<div class="section-badge">
<i data-lucide="users"></i>
<span>Who It's For</span>
</div>
<h2 class="section-title">Built for Teams That Run Their Own Infrastructure</h2>
</div>
<div class="features-grid">
<div class="feature-card">
<div class="feature-icon">
<i data-lucide="layers"></i>
</div>
<h3 class="feature-title">Platform & Infrastructure Teams</h3>
<p class="feature-description">Consolidate more Linux workloads per machine without a hypervisor tax, without noisy neighbors, and without renewing virtualization licenses.</p>
</div>
<div class="feature-card">
<div class="feature-icon">
<i data-lucide="brain"></i>
</div>
<h3 class="feature-title">AI Infrastructure Teams</h3>
<p class="feature-description">Sandbox untrusted agents and generated code in their own kernels, without nested virtualization. Large-memory instances are ready in about 110 ms.</p>
</div>
<div class="feature-card">
<div class="feature-icon">
<i data-lucide="heart-pulse"></i>
</div>
<h3 class="feature-title">SRE & Fleet Operations</h3>
<p class="feature-description">Apply kernel security patches across the fleet with zero downtime and instant rollback. A crash in one kernel stays in that kernel. No maintenance windows required.</p>
</div>
</div>
</div>
</section>
<!-- Products -->
<section id="products">
<div class="section-container">
<div class="section-header">
<div class="section-badge">
<i data-lucide="package"></i>
<span>Products</span>
</div>
<h2 class="section-title">Built on Split-Kernel Architecture</h2>
<p class="section-subtitle">Three open-source software products built on split-kernel architecture. Download from GitHub, deploy on your infrastructure.</p>
</div>
<div class="cards-grid">
<div class="capability-card primary">
<div class="capability-icon">
<i data-lucide="layers"></i>
</div>
<h3 class="capability-title">Multikernel Private Cloud</h3>
<p style="font-family: var(--font-mono); font-size: var(--text-xs); color: var(--accent-700); margin-bottom: var(--space-3); letter-spacing: 0.04em;">Private cloud operating system</p>
<p class="capability-description">Private cloud without the hypervisor. Consolidate Linux workloads on bare metal in your data center, or bring your own cloud: the VMs you already rent. Each gets its own kernel, dedicated cores, and native hardware access.</p>
<div class="capability-features">
<div class="feature-item">
<i data-lucide="check"></i>
<span>Native performance on hardware you own or VMs you rent</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>Zero noisy-neighbor interference</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>No hypervisor overhead</span>
</div>
</div>
<div class="value-cta" style="margin-top: var(--space-4);">
<a href="/products/cloud.html" class="btn-link">
Learn more
<i data-lucide="arrow-right"></i>
</a>
</div>
</div>
<div class="capability-card">
<div class="capability-icon">
<i data-lucide="brain"></i>
</div>
<h3 class="capability-title">Multikernel Sandbox</h3>
<p style="font-family: var(--font-mono); font-size: var(--text-xs); color: var(--accent-700); margin-bottom: var(--space-3); letter-spacing: 0.04em;">AI agent sandboxing runtime</p>
<p class="capability-description">Kernel-level sandboxing for AI agents. Each agent runs in its own kernel on dedicated cores, with strong isolation. No nested virtualization required.</p>
<div class="capability-features">
<div class="feature-item">
<i data-lucide="check"></i>
<span>32 GiB instances ready in 111 ms</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>Kernel-level agent isolation</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>Zero-copy shared model weights</span>
</div>
</div>
<div class="value-cta" style="margin-top: var(--space-4);">
<a href="/products/sandbox.html" class="btn-link">
Learn more
<i data-lucide="arrow-right"></i>
</a>
</div>
</div>
<div class="capability-card">
<div class="capability-icon">
<i data-lucide="heart-pulse"></i>
</div>
<h3 class="capability-title">Multikernel LiveUpdate</h3>
<p style="font-family: var(--font-mono); font-size: var(--text-xs); color: var(--accent-700); margin-bottom: var(--space-3); letter-spacing: 0.04em;">Zero-downtime kernel upgrade tool</p>
<p class="capability-description">Zero-downtime kernel upgrades with instant rollback. Apply security patches and kernel updates without rebooting or interrupting running workloads.</p>
<div class="capability-features">
<div class="feature-item">
<i data-lucide="check"></i>
<span>Zero-downtime kernel upgrades</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>Memory survives app-kernel crashes</span>
</div>
<div class="feature-item">
<i data-lucide="check"></i>
<span>Instant rollback</span>
</div>
</div>
<div class="value-cta" style="margin-top: var(--space-4);">
<a href="/products/liveupdate.html" class="btn-link">
Learn more
<i data-lucide="arrow-right"></i>
</a>
</div>
</div>
</div>
</div>
</section>
<!-- Closing CTA -->
<section id="get-started" class="value-section">
<div class="section-container">
<div class="section-header">
<h2 class="section-title">Run Your Infrastructure on Kernels, Not Hypervisors</h2>
<p class="section-subtitle">Multikernel is open source and runs on the hardware you already have. Start with the docs, or talk to the team that builds it.</p>
</div>
<div class="button-group">
<a href="/getting-started.html" class="btn btn-primary">
<i data-lucide="rocket"></i>
Get Started
</a>
<a href="https://calendar.app.google/nc1upkqQoUeoukdD9" target="_blank" rel="noopener noreferrer" class="btn-link">
Schedule a Demo
<i data-lucide="arrow-right"></i>
</a>
</div>
</div>
</section>
</main>