home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!slc6!trier
- From: trier@slc6.ins.cwru.edu (Stephen C. Trier)
- Newsgroups: comp.protocols.tcp-ip
- Subject: Re: BootP experience
- Date: 29 Dec 1992 19:03:18 GMT
- Organization: Case Western Reserve University, Cleveland OH (USA)
- Lines: 90
- Message-ID: <1hq7dmINN2qj@usenet.INS.CWRU.Edu>
- References: <smiller.725144237@qualcom>
- NNTP-Posting-Host: slc6.ins.cwru.edu
-
- Hello, Scott!
-
- You write:
- >It seems to work great, but I'd like some feedback from someone using
- >this in the real world managing 1000+ machines.
-
- We fit, then. The last time I checked, we had 3000 machines on our net,
- of which about 2500 use BOOTP for configuration. About 2/3 of those are
- MS-DOS, and the rest are Macs. Probably about a dozen systems use BOOTP
- for booting; the rest use it strictly for IP configuration.
-
- >Some of my questions include: How many servers per x nodes/nets?
-
- The standard Unix BOOTP servers use hash tables for their addresses, so
- they scale surprigingly well. We have two main BOOTP servers, for
- redundancy, with 4200 lines in the BOOTP table. (No, I don't know why
- there are so many entries for only 3000 hosts. Either there's some
- redundancy in there, someone hasn't been deleting retired network cards,
- or my size numbers are out-of-date.)
-
- By the way, we include all known systems in our BOOTP table, not just
- those that need BOOTP. You might want to consider this -- this turned
- out to be much more convenient than I expected.
-
- >Has anyone modified bootp to dynamically assign IP addresses to portable
- >(palm-/lap-tops) machines?
-
- We treat portables just like normal machines, but we can get away with
- that because we are not subnetted. I don't know how you'd deal with them
- on a routed network.
-
- I have heard of dynamic RARP servers. I do not know whether dynamic
- BOOTP servers exist.
-
- >Have any security issues arisen from using BootP?
-
- This has crossed my mind. Some denial of service attacks are possible
- when using BOOTP for config info, but they seem not to be too severe.
- I haven't been able to come up with major problems in it when it is used
- only for bland config info.
-
- BOOTP-to-boot opens up a whole batch of possible holes, because the boot
- kernel is then up for grabs -- anyone who feels like running BOOTP and
- TFTP servers can control what kernel the booting machine gets. That
- obviously opens a large can of worms.
-
- Since we use BOOTP pretty much only for config purposes, this latter hole
- is not a problem. The weakness is also common to most diskless-booting
- schemes, not just BOOTP.
-
- The thing to remember is that BOOTP should not change your trust decisions,
- because there is no way to guarantee that a supposedly BOOTP-configured
- machine is what it says it is.
-
- >How much of an impact does using BootP have on a network in terms of
- >broadcasts/second?
-
- The impact is utterly miniscule. I can't get hard numbers for you because
- of the date (much of the campus is not here this week), but I can assure
- you that we see far more traffic from ARP than from BOOTP.
-
- You can compute some rough numbers yourself. Assume 1000 computers, all
- getting restarted once an hour, 24 hours a day, and assuming that the BOOTP
- server always answers their requests on the first try, you should be seeing
- one BOOTP broadcast every 3.6 seconds.
-
- These numbers are tuned for worst-case. Around here, we see one or two
- BOOTP broadcasts a minute.
-
- If you are really interested in some hard numbers, drop me a note around
- January 20, the beginning of the new semester, and I'll grab a network
- analyzer trace to see what kind of numbers we get. I could also get you
- some real numbers on how many machines run BOOTP.
-
- >Is there anything I should be aware of before I start migrating the entire
- >company to use BootP?
-
- Not that I can think of. BOOTP is an incredible boon for large networks.
- It saves _immense_ headaches. We have a base software package that everyone
- gets when they get a network card. (They can add on whatever specialized
- packages they need.) Once the network card driver is installed on the
- machine, *no* further customization is necessary. People can delete their
- software and replace it with a copy of their neighbors, and no address
- conflicts will result. It saves a lot of hassle.
-
- --
- Stephen Trier "We want to offer you a price that you
- Network software type just can't afford to take advantage of."
- Case Western Reserve University - Sales blurb from HSC Software
- trier@ins.cwru.edu
-