Unikernel Platform

Truespar Silicon

Run Truespar products as lightweight virtual machines. No operating system, no containers. A single binary boots directly on a KVM hypervisor and starts serving in under 100 milliseconds.

84 ms Boot Time

From power-on to serving queries. No kernel to decompress, no initramfs to unpack, no systemd to run. The path from boot to your database is a straight line through hardware initialization.

Hardware Isolation

Each instance runs inside its own KVM virtual machine with dedicated CPUs, memory, and block devices. The isolation boundary is the CPU itself, not software namespaces.

~10 MB Binary

The entire VM - database engine, networking stack, filesystem, TLS - fits in a single compressed binary. 10-20x higher density than containers on the same hardware.

The Database Is the VM

Traditional deployments stack layers: hardware, hypervisor, operating system, container runtime, container image, and finally the database. Each layer adds latency, memory overhead, and attack surface. Silicon removes all of them except the hypervisor.

Silicon is written entirely in Rust, from the boot code to the database engine. There is no C library, no POSIX runtime, no third-party kernel. Every line of code that runs inside the VM was written for this purpose.

Why Not Containers?

Containers share the host kernel. Every container on a machine runs inside the same Linux kernel, separated by namespaces and cgroups - software boundaries that have been repeatedly bypassed by escape vulnerabilities. A single kernel exploit compromises every container on the host.

Silicon runs inside a hardware-enforced VM boundary. Each instance has its own virtual CPUs, its own memory, and its own block device. There is no shared kernel to escape from.

ContainerSilicon
OS kernelShared Linux kernelNone
System librariesglibc, openssl, etc.None (statically linked)
Shell / SSH/bin/sh, bash, opensshNone
Package managerapt, dpkgNone
Process tableMultiple PIDsSingle process
Container runtimecontainerd + runcNone
FilesystemOverlayFS layersSiliconFS (native)
Image size200-500 MB~10 MB

Performance Without a Kernel

In a container, every network packet and every disk write crosses the kernel boundary via a system call. Each call costs roughly 100 nanoseconds - a pipeline flush, a privilege transition, a TLB invalidation. Silicon eliminates system calls entirely. I/O and memory allocation are function calls within a single address space.

OperationContainer (Linux)Silicon
System call overhead~100 ns per call0 ns (function call)
Network packet1-5 µs (kernel TCP)100-500 ns (direct virtio)
Disk write10-50 µs (VFS + block layer)1-5 µs (direct virtio)
Context switchConstant (kernel scheduler)Zero (single address space)

Boot Sequence

Silicon boots in under 100 milliseconds - from power-on to accepting queries. The kernel initializes in 15 milliseconds: CPU setup, memory manager, heap, interrupt controllers, and the scheduler.

PhaseTime
Kernel initialization (GDT, IDT, memory, heap, scheduler)15 ms
PCI discovery + disk mount + SMP boot (4 cores)42 ms
Application startup (scan databases, ready to serve)27 ms
Total: power-on to serving queries84 ms

For comparison, a container typically takes 2-5 seconds. A traditional VM with a full OS takes 10-30 seconds. Silicon enables instant scaling and instant recovery - a new instance is serving queries before a container would have finished pulling its image layers.

Hypervisor Support

Cloud Hypervisor

Fully supported. PVH boot, ACPI graceful shutdown, multi-queue virtio-net and virtio-blk, TSC deadline timer, up to 64 vCPUs and 512 GB RAM.

Firecracker

Fully supported on v1.15.0+ with PCI. The same binary adapts automatically - ECAM base, shutdown method, RTC, and PCI BARs are all detected at boot.

Zero Configuration

No environment variables, no config files, no hypervisor detection flags. The kernel probes hardware at boot and adjusts automatically. One binary, any supported hypervisor.

Available Products

Traverse Silicon

The Traverse graph database running as a unikernel. Full Bolt protocol, HTTP API, WebSocket, and gRPC - all in a ~10 MB binary that boots in 84 ms.

Get Started

Learn More

Dive into the internals - boot sequence, scheduler, filesystem, networking, and memory management.