cmcmanis@Eng.Sun.COM (Chuck McManis) (07/14/90)
There is something weird here and for the life of me I can't figure out what is really going on. It certainly looks bad, but I'd like to have the "full story" before I make any conclusions. Item. Kerberos V5 was accepted as part of the OSF DCE offerring with HP/Apollo changes. Item. The RFT process required that all submitted technology be "running" on June 1rst. Which makes sense because of their schedule to release DCE. Item. Presumably OSF has had something running that they are calling Kerberos since the 1st of June. Joe Pato made the comment that HP/Apollo would probably keep their changes as a proprietary part of the OSF offerring (which is perfectly reasonable since it no doubt adds value). Questions : Where is the core part of this V5 Kerberos? (The V5 protocol engine) Will V5 be OSF proprietary or will there be continue to be an MIT open version? Why can't we get any information on version 5? (Even the final specification.) Where is the V5 alpha so that people can make sure they aren't depending on V4 features that may vanish? If Kerberos didn't make the 1 June deadline for OSF does that mean it is out of DCE? Mumblings : From the outside looking in, the appearance of a conspiracy to keep Kerberos out of the hands of people who might ship it before OSF is evident. I don't think anyone is that silly so I discourage such rumors whenever I hear them. What I would like is that the people who keep talking about V5 in the present tense would let someone else see it. How about a "final" spec even? Confused in Mountain View,
jis@MIT.EDU (Jeffrey I. Schiller) (07/18/90)
Date: Fri, 13 Jul 90 13:34:46 PDT From: cmcmanis@Eng.Sun.COM (Chuck McManis) There is something weird here and for the life of me I can't figure out what is really going on. It certainly looks bad, but I'd like to have the "full story" before I make any conclusions. I'll try to fill you in on the MIT view of all this. I don't think anything weird is going on. What you see is the result of an interesting chain of events. As you know MIT makes Kerberos available in a "public" sense. We freely distribute it via FTP with a copyright that allows others to make use of the technology and the distributed code itself. One of the implication of this is that others, including vendors can pick up our work and incorporate it into their products. In general we encourage this as a "good thing" for it leads to industry acceptance of what we believe to be good technology. As part of encouraging wider acceptance and use of Kerberos, we at MIT encouraged discussion, a lot of it via this mailing list, on how to improve kerberos. The currently distributed version, version 4, was designed and implemented to solve the specific authentication problems that needed addressing for Project Athena. As such it does not support desirable features that were not in the original problem space of Project Athena (for example non-IP networking technology or complicated namely strategies). The result of this discussion process was the design of V5 kerberos. V5 eliminates many of the protocol obstacles inherent in V4 and provides notable enhancements (proxy tickets and renewable tickets just to name two). As I am sure you are also aware OSF has a technology solicitation process. HP/Apollo responded to this solicitation with a technology that was based on Kerberos with proprietary extensions (which is OK in that V5 was designed to allow such extensions). This is also a legitimate thing to do given the way that we have copyrighted kerberos. MIT did not play a direct role in the proposal from HP/Apollo to OSF. We did encourage all concerned to submit V5 based technology rather then V4 technology which we would like to make obsolete in the not so distant future. HP/Apollo was happy to do this and OSF was happy to accept this from what I can see. That is where we stand today vis a vis the OSF DCE offering. Obviously there is some confusion as to who provides the "reference" kerberos. Is it MIT or OSF. This is an interesting issue. To help clarify this and reduce confusion (sometimes our own confusion!) we have been having conversations with OSF. These discussions are currently ongoing and at a point where it is hard to say what the future holds. However there are some things that we can state pretty clearly. It is not MIT's intent to make Kerberos (any version) proprietary technology. When it is ready, MIT will make V5 kerberos available in the same fashion as we make V4 available today. So Where is V5 you ask? Behind schedule is the answer. To date we have had very limited resources for the development of kerberos (after V4 was finished). The primary developers at MIT are pretty busy people who all have other responsibilities that have precluded an intensive effort to get V5 ready. John Kohl, a DEC employee assigned to Project Athena, has been doing a yeoman's job of working on V5, more or less on his own. Where's the final spec you ask... Well that's my fault. I volunteered to make the last major round of changes to the V5DRAFT RFC (version 2) which is more or less a rework to remove the bit level packet formats and convert them to an ASN.1 specification (which is what John is implementing). I have been busy with other things and haven't had time to work on it. Now the good news. We will soon be putting more manpower and internal priority on V5 kerberos. Although I cannot commit to dates at this point, I am confident that the rate of progress on getting the code in alpha-test shape will increase. Let me address your questions one at a time. Where is the core part of this V5 Kerberos? (The V5 protocol engine) Being worked on. I think the text of my message makes this clear(er). Will V5 be OSF proprietary or will there be continue to be an MIT open version? As I stated above, MIT intends for V5 to be as available as V4 from MIT. Why can't we get any information on version 5? (Even the final specification.) I consider myself poked! Where is the V5 alpha so that people can make sure they aren't depending on V4 features that may vanish? I don't know of any V4 features that will vanish... but your point is taken. The short answer is: We're working on it. If Kerberos didn't make the 1 June deadline for OSF does that mean it is out of DCE? I don't know. I cannot speak for OSF, nor in fact do I know. Mumblings : From the outside looking in, the appearance of a conspiracy to keep Kerberos out of the hands of people who might ship it before OSF is evident. I don't think anyone is that silly so I discourage such rumors whenever I hear them. What I would like is that the people who keep talking about V5 in the present tense would let someone else see it. How about a "final" spec even? I hope I have clarified things for you (and others). From what I can see from MIT, there are no conspiracies hiding, just some honest people working hard. Confused in Mountain View, Overworked in Cambridge (note the time on this message header, I've been at work since 8:30AM). I hope this diatribe helps. If you have any questions feel free to ask, either via e-mail (to me and cc'd to the list if you like) or via telephone at (617) 253-8400 (please don't everyone call at once!). -Jeff