Home Multikernel Private Cloud Multikernel Sandbox Multikernel LiveUpdate ARM Platform RISC-V Platform OEM & Embedded SaaS & Database Clouds High-Frequency Trading Technology How It Works Performance FAQ Getting Started Blog About 中文 GitHub Schedule a Demo
Multikernel LiveUpdate

Upgrade Your Kernel Without Rebooting

Apply kernel security patches and updates without downtime. Boot the new kernel alongside the old one, migrate workloads, and roll back instantly if anything goes wrong. No maintenance windows required. Available standalone as a per-node subscription: adopt it without changing the rest of your stack.

Zero-Downtime Kernel Upgrades

Kernel upgrades are one of the most painful operations in production Linux. Teams delay security patches for weeks because they require reboots. Livepatching services work around it by patching individual functions in place. LiveUpdate replaces the entire kernel, and the old one keeps running until you commit.

Boot New Alongside Old

Launch the new kernel alongside the running one. Both kernels operate simultaneously on separate cores. No downtime window to schedule.

Migrate Workloads

Move applications from the old kernel to the new one gradually. Validate each migration before proceeding. Keep full control over the transition.

Instant Rollback

If something goes wrong, roll back instantly. The old kernel is still running. No reboot, no recovery procedure, no data loss.

Surviving Kernel Crashes

A kernel crash today means a full reboot, minutes of downtime, and every cache starting cold. With Multikernel, a crash stays inside one kernel, and its memory outlives it.

0 Bytes lost across four kernel panics 4 of 4 panics verified intact

An app-kernel panicked mid-write four times. Each time, its memory on DAXFS was intact down to a half-written page, and the rest of the machine kept running. Recovery was a kernel boot, with no log to replay.

Crash recovery

Crash Containment

A panic in one kernel stays in that kernel. Workloads in the other kernels keep running, with no machine reboot.

Warm Restart

The crashed workload restarts on a fresh kernel and maps its DAXFS memory back, so caches and indexes do not rebuild from cold.

Benefits

Ship Security Patches Faster

Kernel updates no longer wait for reboot windows. Apply patches immediately with zero application downtime.

Contain Kernel Failures

A crash in one kernel stays in that kernel. Other workloads keep running, and the crashed workload restarts with its memory intact instead of rebuilding from cold.

Reduce Operator Burden

Eliminates scheduled maintenance windows for kernel updates. Upgrades, validation, and rollback happen while workloads keep running.

How to Get It

Multikernel LiveUpdate is open-source software you install on your own infrastructure. Free to download and build; the enterprise subscription is priced per node per year and buyable standalone.

Open Source

Download and build from source. Perform zero-downtime kernel upgrades on any multikernel-enabled system. Community support via GitHub issues.

Enterprise

A standalone per-node annual subscription for production environments where downtime is not an option.

Validated kernel update paths tested against your workloads
Automated upgrade orchestration with rollback policies
Security patch delivery within 48 hours of upstream disclosure
A ready-to-apply LiveUpdate image for every erratum in your update stream
Priority engineering support with SLA-backed response times
Crash recovery and disaster recovery planning
Quarterly kernel health reviews and upgrade roadmap