The SPEEDOS Project
At that point Keedy decided to design a new operating system and together with his PhD student Klaus Espenlaub at the University of Ulm designed an initial version of SPEEDOS. This was an early attempt to build a SPEEDOS system on conventional hardware. It was important to clarify some new ideas and was an excellent thesis, but unfortunately it lacked the hardware needed to provide an efficient running system, and at the same time it left open some very difficult practical questions (e.g. how to translate 256-bit virtual addresses in a cost-effective and efficient manner and what was involved in practice to implement solutions for the confinement problem). A copy of Espenlaub's PhD (which is in English) is available under Downloads.
In 2005 Keedy retired and decided to spend the time which he could allot to SPEEDOS solving these open issues and at the same time clarify for others the main principles governing how a SPEEDOS System could appear. He has now prepared a manuscript in textbook style ("Making Computers Secure, volume 1") which explains the choices that are relevant for the design of a secure system (in the computer architectural meaning of security). This can be downloaded below.
Building on this first volume he has also written a much more detailed text in a more technical style ("Making Computers Secure, volume 2") which explains how a final SPEEDOS implementation can be realised in detail, including how the kernel functions, how 256-bit SPEEDOS virtual addresses can be efficiently translated into main memory addresses using conventional address-translation hardware, how qualifying modules can be implemented (and therefore how the confinement problem can be solved), how SPEEDOS can be networked, including over the Internet, etc. It also indicates how operating systems can be built on top of the SPEEDOS kernel, and it provides a detailed example of a banking system. Volume 2 can also be downloaded below.
I write "the time which he can allot to SPEEDOS" because I also have other projects that were not completed before my retirement. These include the design of a new programming language named Timor, which is a persistent language that supports the information-hiding principle and is ideal for developing both an actual SPEEDOS system and also application programs for SPEEDOS. A description of the Timor language design has now been completed and can be downloaded from the Timor website (see https://www.timor-programming.org/) and from this site under Downloads. Timor could be used to program a final SPEEDOS implementation and application programs for it.
We now consider the key features of the SPEEDOS design.