The entire KaOS hypervisor is embedded within the Linux 2.6 kernel.
Simply drop in a KaOS-enabled Linux kernel and reboot...
Carbon Mountain takes advantage of a feature in the Linux 2.6 kernel that allows a compressed cpio image
to be embedded within the kernel. The kernel decompresses the cpio image into memory during the boot process.
The small footprint of the KaOS hypervisor platform enables the entire system to be embedded with the kernel.
The operating system in KaOS is literately attached to the kernel, hence the name Kernel Attached Operating System
or KaOS. The entire solution is contained within a small kernel image, typically under 10MB in size.
Integrated Image
Integrated images are widely used in enterprise grade routers, switches and other networking appliances. These
images are typically referred to as firmware images and run from memory. Integrated images are easier to
quality control and reproduce problems. They do not suffer from unknown variables such as dependency problems, typos
in critical configuration files or intermittant storage problems. KaOS brings the same mission critical solution to
the Hypervisor.
Zero Installation
Simply booting a KaOS-enabled kernel provides an instant hypervisor platform. KaOS can be booted any way a Linux kernel
can be booted - PXE, USB Flash, Local Disk, Removable Media etc. An existing Linux system simply needs to be rebooted with
a KaOS-enabled kernel. Upgrading from Fedora, Ubuntu, and other Linux distributions is just a CTRL-ALT-DELETE away.
Small Footprint
Unlike traditional Linux platforms that are trying to slim down to become a better fit in the cloud and virtualization focused market.
KaOS is specifically designed to be light-weight, secure and highly available. Compared to KaOS and vKaOS, most Just Enough
Operating System or JeOS platforms, are a good effort but are still Just Too Much Operating System.
The KaOS footprint, including the Linux kernel is under 10MB. Even uncompressed, the KaOS platform is still a fraction of the size of
the Ubuntu Server JeOS solution. The vKaOS platform is even smaller than KaOS. The vKaOS and AppQueue module framework do not pull in extra
software dependencies as package management based Linux distributions such as Fedora / Red Hat, Ubuntu, OpenSuSE etc do. This smaller footprint
results in smaller images, which are faster to move from Hypervisor to Hypervisor, or from Data Center to Data Center, or even Cloud to Cloud.
Only the required libraries and binaries are installed, reducing the overall security and software failure risk of the system. KaOS further
reduces space and enhances security by selectively compressing libraries and binaries in read only filesystems such as SquashFS.
Optimized
The typical KaOS-enabled Linux kernel has kernel modules disabled, eliminating an entire attack vector from the system. The KaOS SDK provides
a hardware profile system, that allows IT professionals to customize and build Linux kernels for their specific hardware. Once a hardware
profile has been built, it can be used on all the servers that match that hardware profile within an organization. Carbon Mountain provides a
free web service called KaOS Build that makes it easy for IT professionals to configure, download and
share hardware profiles for KaOS. By selecting just the drivers that are needed by the server, the system boots a lot faster and has less drivers
running, reducing the risk of hitting zero-day bugs.
Fault Tolerant
Since KaOS is a single image that runs from within memory, as long as the system can boot the image, the platform will be available. KaOS is
designed to be highly available, based on years of experience by engineers who managed servers often on different continents. If local storage
is unusable, KaOS can continue to run using network storage options such as iSCSI or in a reduced capacity mode with ramdisks. The node will
continue to operate even in the event of a complete local storage failure.
Traditional Linux distributions by comparison can become completely unusable when local storage problems occur. File system check failures often
drop servers to maintenance mode, requiring console access. Sometimes disk problems will result in the root filesystem becoming read-only, logging
will fail and services will fail to start due to existing pid files that cannot be removed. All of these problems, while not everyday occurances,
will result in lengthy downtime.
Click here to find out how KaOS is used to build next-generation data center environments.