[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

a sad day for microkernel research (Jochen Liedtke)

From: Roland McGrath
Subject: a sad day for microkernel research (Jochen Liedtke)
Date: Mon, 11 Jun 2001 16:39:20 -0400 (EDT)

To those it may concern, I've just found out that Jochen Liedtke died
yesterday.  He was the architect of L4 and author of important papers on
microkernels.  I only met him once briefly at a conference, but he was a
fine asset to the brotherhood of operating systems research.

> It is with sorrow that we announce the death of Jochen Liedtke. He
> died unexpectedly on Sunday June 10th.
> The funeral ceremony will take place on Friday, 15 June 2001, 3pm
> in Obergrombach near Karlsruhe, Germany.
> Jochen Liedtke, in memoriam.
> Jochen studied mathematics at the University of Bielefeld, completing
> his diploma in 1977. The focus of his thesis was the novel programming
> language ELAN.  Jochen's first operating system was a by-product; a run
> time environment for ELAN was needed on small Z80 micro computers. The
> result of Jochen's effort, EUMEL, was based on two simple principles:
> persistent processes and data spaces. All data of the entire system
> including process control blocks and data space descriptors were
> contained in these data spaces. They could be copied efficiently and
> atomically using copy on write and garbage collection techniques. By
> copying the "data space of data spaces" every few minutes, a complete
> copy of the entire system state was taken and lazily written out to
> disk. Thus, process persistence came for free (at least conceptually).
> Sending around data spaces in synchronous messages was the only means of
> process interaction which made it easy to build a simple distributed
> EUMEL system.  The paging device was a floppy disk (what else on a cheap
> computer at that time).
> In 1984 Jochen moved to GMD, the German National Research Center to
> build a "native code" version of EUMEL, called L3. This was the time
> when microkernel based systems were en vogue. Soon however, many
> researchers gave up their attempts to build the really fast message
> passing systems that were needed to run device drivers and other
> performance critical components at user level. Declaring "The Increasing
> Irrelevance of IPC Performance for Microkernel-Based Operating Systems",
> IPC was avoided by co-locating drivers and other components back into
> the kernel. Jochen however accelerated IPC by factor of 20 over
> comparable systems. The methods, most based on a thorough understanding
> of the interaction of modern microprocessor architectures with operating
> systems, were published in his SOSP 1993 publication ("Improving IPC by
> Kernel Design"). Still, L3 was not widely used (except in about 3000
> installations in German law practices) because of its very special user
> and programming interface. Consequently, Jochen started close
> cooperation with Dresden University's operating systems group to build a
> Unix-like interface on top of L3. During this time, he invented
> hierarchical external pagers, another important feature that allowed
> physical memory management to be done in user-level pagers (SOSP 1995,
> "On Micro-Kernel Construction").
> Jochen completed his PhD on guarded page tables in 1996 at TU Berlin. He
> demonstrated that not only was he a successful operating systems
> builder, but he also excelled with contributions to computer
> architecture.
> Jochen began working at IBM's T.J. Watson Research Center, NY in 1996.
> The result was L4, a 12KB extremely fast "second generation"
> microkernel. During many visits to Dresden's operating systems group, he
> helped to build L4Linux, a user-level implementation of the Linux kernel
> that demonstrated the effectiveness of Jochen's approach (SOSP 97, "The
> Performance of microkernel-based Systems").  His work gave research on
> microkernel systems fresh impetus and gained him international
> acknowledgement.
> Since 1999, he became professor for operating systems at the University
> of Karlsruhe.  He was an inspirational professor, adept at keeping
> students deeply interested in his subjects. This was reflected by the
> high popularity of his lectures among students.
> Jochen also continued working on SawMill Linux, a multi-server version
> of Linux, and other related micro-kernel subjects. Several operating
> system research groups in Europe, Australia and the US either base their
> work on Jochen's or draw from his results. He was highly acknowledged
> for his scientific achievements which are reflected in numerous
> honourable appointments to program committees.
> Those who knew him remember the energy, stamina, and astute analysis
> with which he not only tackled his own scientific issues but also
> supported staff and colleagues. Not only expertise, but also
> friendship and mutual understanding played an important role between
> him and those he worked with. This cooperation developed numerous
> friendships over many years and large distances.  Many colleagues,
> staff and students enjoyed his generous hospitality, exceptional
> culinary skills, and taste for good wine.
> Jochen is survived by his wife Adelheid with whom he lived together in
> harmony. Her constant support was instrumental in his many
> achievements.
> Hermann Härtig
> Dresden University of Technology
> Kevin Elphinstone
> University of Karlsruhe

reply via email to

[Prev in Thread] Current Thread [Next in Thread]