NetKernel's foundation is a robust microkernel that efficiently schedules every resource request. The minute size of the microkernel contrasts with the power of the abstraction it embodies.

NetKernel is extremely efficient with memory.

A minimal deployment runs in only 4 MB of heap. The complete NetKernel Standard Edition operates in 12 MB. Enterprise application can run in 64 MB. These figures are especially remarkable when compared with typical J2EE server deployments.

NetKernel's microkernel scales from single-core processors to SMP multi-processor, multi-core configurations. The asynchronous scheduler guarantees threads are allocated optimally to requests and minimizes native-thread context switching.

All resource request results can be cached using the resource address as the cache key. With the microkernel's fine grained control over resources and threads, this results in astonishing performance gains, even in constrained memory environments. The practical result is dynamic applications with static performance.

High availability is vital. Because everything in NetKernel is linked via logical addresses only at the moment of a request the microkernel can switch references between entire modules while a system is running. This capability allows NetKernel to offer 24-7 operation and live updating of a system. If an update turns out to be incorrect, instantly rollback the configuration to a previous state.

To meet our stringent requirements, NetKernel takes control of its inbound requests using a request throttle. This throttle shapes the request profile for optimal handling by each deployment configuration. Overloads are handled gracefully while application service-levels are maintained.

NetKernel supports the active: URI scheme that unifies service activation and resource requests. Everything in NetKernel is either a service or a resource. This simple and uniform view of software opens powerful possibilities for system design and implementation.