steve@yoyodyne.mit.edu (Steve Jones) (02/16/91)
Greetings, A while back I posted a request to this newsgroup for help in getting information about the Sprite project at UC Berkeley. I was shocked at the large number of helpful responses I received, and would like to thank everyone who took the time to respond. I would also like to apologize for taking so long to write this summary, but minor things like a departmental reorganization have gotten in the way. ;-) Where to start. First, there are several papers available for anonymous FTP on sprite.berkeley.edu. Please note that this is just an alias for the machine allspice.berkeley.edu - don't be surprised when you are greeted by allspice. There are only a few papers, and they aren't kept in any particular place, and there are some papers on topics that are tangentally related to Sprite, such as Tcl and Tk (see comp.windows.x for more info on these - or get the papers...). Here are some of the papers available that were most relevant to the OS itself: amsp.ps.Z 4-Feb-1991 draft comparing Sprite vs. Amoeba Douglis, Kaashock, Tanenbaum, Ousterhout lfsUsenix90.ps Summer '90 USENIX paper in the Log-structured File System Compares lfs/Sprite vs. SunOS on comparable hardware Rosenblum, Ousterhout mig.ps paper on Sprite process migration (no date) Douglis, Ousterhout sedmsII-test.ps March '91 SEDMS-II paper on cache performance in Sprite Welch In addition to these, some reports are available based on work that's been done at Digital's Western Research Lab, a.k.a DECWRL. These papers can be retrieved from the archive server run by that organization. Papers are retrieved by referencing the year and report number for a given article. They are only available in Postscript form. To get them directly from the source, send the following two lines to "wrl-techreports@decwrl.dec.com" send postscript 89/5 send postscript 90/4 These papers discuss the following issues: 89/5 An implementation of NFS using Sprite cache consistency protocol Srinivasan, Mogul 90/4 The performance and trade-offs of Sprite's dynamic virtual memory vs. file system cahce division of physical memory. Nelson An information packet has been produced by the Sprite group and may be requested by sending an electronic mail message to the address "sprite-request@sprite.berkeley.edu" making sure to reference the "sprite packet." In addition to these ways to procure the actual papers, I received a LaTeX bibliography of Sprite documents courtesy Jim Mott-Smith of the Sprite group. I believe this can also be requested from the "sprite-request" address, but don't quote me on that. I've taken all of these publications and made them available for anonymous FTP on yoyodyne.mit.edu. If you can't get anonyFTP to work, send me some EMail. Note, this includes the bibliography, but not necessarily the papers in the bibliography. yoyodyne.mit.edu:~ftp/pub/info/Sprite/ [18.80.0.200] Many thanks to all the people who responded: Fred Douglis Adam Glass Mike Kupfer Jeffrey Mogul Jim Mott-Smith Craig Partridge Brian Pawlowski Rich Scott Brent Welch If I left anyone out, I will feel deeply ashamed. So don't tell me ;-) Also be aware that I have no affiliation that I know of with UCB or DECWRL, and what I've done may be completely against their wishes. If I am informed that they prefer I not do any of this specifically or generally, the offending act shall be remedied swiftly. Hope this is of use to someone. --Steve. -- Steve Jones MIT/VAX Resource Center, Cambridge MA 02139 617/253-7438 Internet addresses: steve@isis.mit.edu steve@yoyodyne.mit.edu "Chaos will ensue if the variable i is altered..." - SysV Programmers Guide