Going up against the new wave of desktop Unixes, Golden, Colorado-based System Six Inc, the company that has gone to Russia for its answer to Unix (CI No 2,200) intends to focus initially on Intel Corp hardware supporting AT, EISA, C-bus and Multibus configurations. A RISC version is planned to follow six months after initial release. The software is compliant with IEEE Posix P.1003.4, 1003.6, 1003.7 and 1003.P. The Russians have developed their own MS-DOS translator, Dosix, that operates independent of X Window and goes into beta test this month. It is claimed to enable Windows and MS-DOS applications to run on the console but it will not yet make them network-extensible. TCP/IP support is due before autumn, threads and Network File System support will follow. System Six Inc’s director of software engineering, Brad Erlwein, formerly with Zenith Data Systems Inc, says there are still minor holes in Usix that have to be plugged and it is Erlwein’s job to see that the software is properly tested and has adequate technical support. While the kernel is now stable, he says, the Russians are taking their first crack at a TCP/IP stack on the UDP protocol.

Swedish ex-patriate

It will be autumn before TCP/IP is fully supported. Usix also currently lacks threads and Network File System support. He feels it will be the end of the year before Usix is fully networked with X Window and systems adminstration support and a full complement of Unix-like utilities. But it should be out at user sites in Sweden, South America and the US in the next few weeks. System Six acknowledges that the US, or even the industrial West, may not be its real market. It claims however that the growth of Unix-like systems for the rest of the decade is projected at 5% to 45% in the rest of the world, versus a flatter 15% to 20% in the US. System Six Inc is the brainchild of its president Tomas Jablonski, a Swedish ex-patriate, and its vice-president of research and development Roman Tankelevich, a Russian software expert and Soviet dissident. The two were originally brought together by Jablonski’s work in Sweden on behalf of D-nix, the Diab Data AB Unix variant, supposedly a prototypical microkernel, which is heavily used in L M Ericsson Telefon AB’s telephone systems and eventually bought by the Swedish government. Diab Data is now 75% owned by Compagnie des Machines Bull SA. Jablonski looked to the Russians to get a cost-effective alternative to D-nix and System V.4. It was from that work that the idea for Usix was born in the mid-1980s. Through Tankelevich, groups of Russian scientists were asked to bid on building it, a project that could be done there for tens of millions of dollars less than it could in the West. The bid was won in 1990 by Electronmash, the 65,000-employee Soviet predecessor of INEUM. One of the reasons Jablonski and Tankelevich decided to headquarter System Six in the US was the protection of intellectual property afforded software in the US. System Six says it has around $10m in first round investment from seven international banking and private sources. No venture capitalists are involved and six more investors are said to be in the wings for additional tranches.

By Maureen O’Gara

The Russian INEUM’s Usix was designed from the ground up on Intel Corp 80286 and 80386 machines and the Corollary Inc multiprocessor C-bus around newer technologies than its ageing Unix predecessor and so is said to be object-oriented and microkernel-based, two leading-edge areas that Unix is only now starting to incorporate. Usix’s Russian developers, previously involved with the old Soviet Union’s advanced defence research and members of the interlocking Russian Academy of Science, had in mind to solve the persistent weaknesses in Unix. Usix was implemented from the beginning as a multi-CPU-based kernel. Unix support of multiprocessors, on the other hand, has been only partially successful and done at the expense of portability and performance, System Six says. Usix is said to be fully pre-emptible and supports symmetric multiprocessor architectur

es at all levels from the kernel internals, through device drivers to higher level functions such as TCP/IP. It therefore provides a better, more linear scalability than systems relying on patches to a single-CPU kernel. Erlwein claims the current MP/Usix gets 90% of the power of each of the processors it runs on in a multiprocessor configuration compared with Unix at only 50%. Usix also says that System V.4 is still a typical time-sharing kernel with a few pre-emption windows in the otherwise monolithic critical region. The real-time extensions added to it so far don’t go as far as Usix does. System Six says the Usix kernel can almost always be interrupted. Critical regions are short, few and well-defined. The response time is therefore fully deterministic. Real-time priorities are the priority levels higher than that of the kernel. Processes executing at these priorities can be pre-empted only by processes with an even higher priority. No time sharing of processes executing at the same priority is permitted. The Usix kernel was also designed with the goal of providing the shortest possible interrupt latency, context switching time, semaphore shuffling time and deadlock breaking time. Being microkernel-based, Usix uses servers to provide basic services: file system emulation, serial terminal and console input-output and several other device drivers. Some facilities such as Streams are available both as servers and as linkable kernel modules. Other possible applications include database and screen windowing servers. The Usix file system provides enhanced reliability and user security, System Six says. Several copies of all vital structures are maintained throughout the disk to lower the chance of catastrophic data loss.

Relational database

Directories support additional data needed for higher user security levels. The file system provides total emulation of System V so that older programs that open directories as ordinary files instead of using system calls will function correctly. Usix avoids defining a separate swap file system during installation. An ordinary file is automatically created during installation that will dynamically extend as needed, preventing out-of-swap-area errors and the need to re-install the entire system if the initially allocated swap file system is too small. It also adds three synchronisation levels: full, no and partial. In addition, all the main memory not used by processes or the kernel is automatically used as disk buffers, providing optimised disk input-output performance. Most kernel parameters in Usix can be dynamically changed at almost any time by executing a simple program. This enables the kernel to be reconfigured according to actual demands. System Six says it is also adding several extensions to the basic Usix, including a server-based relational database manager with compatibility modules for users of Oracle, Informix and other packages.