gil@limbic.UUCP (Gil Kloepfer Jr.) (07/07/89)
I am being asked to investigate the possibility of using AIX for a client who demands that software/hardware come from "big blue." Has anyone used AIX yet? If so, how well/poorly does it function? How does it compare to AT&T System V.3? What machines in the workstation/PC hardware class does it support (ie. ATs, 386 (or PS/2), PC/RT)? Are there any specific porting considerations? Does AIX have reasonable graphics support? Answers recommending that we use something else don't count, as we may not have a choice. Please followup to me via e-mail, and I will post a summary if I get sufficient responses. Thanks. Gil. --------- | Gil Kloepfer, Jr. | ICUS Software Systems/Bowne Management Systems (depending on where I am) | ...icus!limbic!gil or gil@icus.islp.ny.us
gil@limbic.UUCP (Gil Kloepfer Jr.) (07/20/89)
Well...I was unprepared for the overwhelming response I got to my article posted a few weeks ago, quoted below. I figured from the response that I got, that it would be only fair that I share all the information I obtained with the rest of the net. My thanks to all who answered -- I'm now much more enlightened than before. Because of the number of responses, I've been forced to edit many of the articles (ie. chop em down) a bit so that they don't make the internet news system sick. If anyone has a problem with this, feel free to write and I will forward anyone's entire response, unless there are objections. As for those who noted problems -- I had a couple of responses from people at IBM research, who might be kind enough to pass along these comments and suggestions to the right development group in the name of making the product better. Anyhow...my article essentially asked... > I am being asked to investigate the possibility of using AIX for a client > who demands that software/hardware come from "big blue." Has anyone > used AIX yet? If so, how well/poorly does it function? How does it compare > to AT&T System V.3? What machines in the workstation/PC hardware class does > it support (ie. ATs, 386 (or PS/2), PC/RT)? Are there any specific > porting considerations? Does AIX have reasonable graphics support? Here is a summary of the responses: |From: sbcs!karish@forel.stanford.edu (Chuck Karish) |Organization: Mindcraft, Inc. | |>If so, how well/poorly does it function? | | Pretty well. IBM seems to be putting a lot of effort into | making it a solid operating system. They're succeeding. | |>How does it compare to AT&T System V.3? | | It's a port of System V.2, with BSD networking utilities | and libraries and some uther utilities added. It uses | HDB UUCP, which works fine. | |>What machines in the workstation/PC hardware class does |>it support (ie. ATs, 386 (or PS/2), PC/RT)? | | The RT and the PS/2 Models 70 and 80 are supported now. | Non-PS/2 386 boxes are not supported. | |>Are there any specific porting considerations? | | Applications that use System V.3 extensions (notably streams) | will require substantial effort to port. Software that | is written to be portable among different versions of | BSD and AT&T UNIX are easy to bring up. You have your | choice of doing it with SysV-style or BSD-style interfaces. | The trend in AIX development seems to be towards supporting | the POSIX and X-Open standards, which include some interfaces | derived from BSD and some from SysV. | |>Does AIX have reasonable graphics support? | | AIX RT-2.2.1 has a solid version of X-Windows v. 11.3. There's | also a stand-alone graphics package, but I haven't used it. | | |I've been using AIX for about two and a half years. It's come a long |way in that time, in terms of both functionality and reliability. Our |machines are used to develop software under contract to IBM, so we |always run up-to-date versions of the OS. | |The documentation is good, but could be better. | |Field service for RT hardware is prompt, and the people know what |they're doing (in Palo Alto, anyway). | |Software support is OK, once you figure out the bureaucracy. | |My only real gripe right now is that some UUCP bugs surfaced in 2.2.1, |and they haven't been fixed yet. The asynchronous line protocols are |not well described in the documentation. | |IBM is doing a clean job of integrating BSD functionality into AIX, |with no universe switches or other such hacks. The stuff's really |there, and you don't have to choose between (in)compatibilities. |From: sbcs!stripes@wam.UMD.EDU | |It runs on PC/RTs, and it runs under VM (IBM mainframe concept), and it runs |on PS/2's. The RT version (at least) has a somewhat broken X-windows, so |it has good graphics support, plus bugs ;-) | |I havn't used AIX mutch, so I can't attest for it's reliability, or it's |feechure list. It seemd to be missing alot of nice BSD stuff that Ultrix |stole... |(sockets for one, or mabie I just couldn't find the right libery to link) |From: well!gors (Gordon Stewart) |Organization: Whole Earth 'Lectronic Link, Sausalito, CA | |As a user and software developer currently using AIX, I would rank |it lowest of all **IXs currently available. I am using 2.2.1 on an |RT (135). | |AIX is based on OLD UNIX, somewhere around System III, when Xenix |split from UNIX -- and has a bunch of SYS V stuff grafted on -- like |the Sys V IPC facilities. | |Many common utilities have names different from the ones you know, |which is a pain. There is no job control (ctrl-Z), and |I get the feeling that the whole thing was thrown togerther in a real |hurry, just to get into the race. |From: uunet!acheron!clarke (Ed Clarke/10240000) |> I work for IBM (Research). | |AIX comes in several flavors. | |AIX 2.x runs on the RT and is based on V.2 rather than V.3. There have |been a lot of BSDisms added - bind 4.8 and sendmail 5.61 compile and run |on my 2.2.1 system at work. There is no bsd job control (yet). NFS is |available ( older version ). | |AIX PS/2 runs on PS2 mod 70 and 80 systems ONLY. Forget about 286 or clone |386 machines ( unless they're exact clones ). Easy to write device drivers |for non standard things ( don't know about console drivers though ). Does |include bsd job control. There's a 'tcf' facility going to be available |sometime in the fall. Sort of interfaces AIX PS2 to AIX/370. I have seen |it run, but don't know anything much about it. No NFS yet. | |AIX/370 runs on 370 type machines ( 309x, 438x etc. ). Seems to work ok |but I don't have much experience with it. Includes some kind of 'tcf'. |NFS is supported. | |All AIX systems support X-windows for graphics ( and tcp as well ). | |Further note: AIX has been given to the OSF and is being used as a base |for their development. You might ask about other vendors products. I |am using the 'motif' window system ( runs on X but is not a product yet ). |Screen looks sort of like the OS/2 presentation manager. |From: lhf@aries5.uwaterloo.ca (Luiz H. deFigueiredo) |> Computer Systems Group at the University of Waterloo | |We in the Computer Systems Group at the University of Waterloo have been |using AIX for some time now on a RT. |Since March, we have also been using AIX on PS2 models 70 and 80 (80386). | |In general, it seems really good for version 1.1 (on the PS2's). | |AIX seems to be fairly standard System V.3 compatible with several touches |of 4.3BSD. | |There's support for X-windows (X11) but it is not great on the PS2 and we |do not have graphics on the RT. | |Please note that the above is *not* the official position of CSG. |I just happen to work here... |From: ames!cs.utexas.edu!uunet.UU.NET!arnor!freimer (Robert Freimer) | |Gil, I'm using AIX on PS/2, RT and /* commented */ machines. So far it's |working fine, I didn't have to reboot it even ONCE. Compare to System V.3? |Well, it's more reliable (though my SysV.3.2 at home also works fine, but |after I had really hard time fixing problems with disk bad block handling, |and I had to end up with another disk drive - it refused to work on that one). |Features - it definitely has more, because it's SysV.2 compatible, but has a |lot of BSD goodies (which I appreciate to use). Porting problems - nothing |special so far. Machines supported - NO ATs, NO AT386's so far, probably |never. Just PS/2 and RT, but I think you'll get more than just NFS. | | Uri. |> gatech!nanovx!msa3b!kevin | |I just installed AIX on a PS/2 model 80. I believe it will run on any of the |'386 micro-channel machines. Since I have JUST installed it, I have only |had time for a cursory examination. It looks pretty similar to SYS V if you |are writing typical applications in C and shell. Note that Korn Shell seems |to be pretty difficult to get for it. One limitation that I have seen that |concerns me is that the "termio" driver only handles up to 9600 baud |(we use modems that will accept data at 19200). | |I have had a LITTLE trouble installing it. | |I will be porting software from an AT&T 3B2/700 running Sys V 3.2.1 |to a PS/2 model 80, running AIX 1.1, so I should get a good picture for |the compatibility in the next few weeks. If you are still interested, |email me in a week or two, and I'll let you know what I've found. |> Bob Izenberg, Brandon Consulting uunet!(attctc,cbis3,elephant)!bei | |A friend just started on an AIX project at IBM and all he uses are RTs. |From: "Stefan M. Vorkoetter" <watmath!watmum!smvorkoetter> |Technical Manager |Waterloo Maple Software | |I am using AIX on an IBM PS/2 Model 70 A21 and I think it is great. It |has a few little tiny bugs, but nothing terrible. Performance is really |good (about the same as our VAX 8650 with one user on it) CPU-wise, and |quite acceptable disk-wise (these 3.5" hard disks just don't cut it compared |to the VAX 8650 disks). | |I am also running DOS Merge to allow me to run various DOS applications, |and it works well, even with "ill-behaved" DOS programs. | |As far as portability is concerned, we just ported the Maple symbolic |algebra package from our VAX to the PS/2. The whole thing, including |running all our diagnostic tests, took one afternoon. In comparison, |our port from VAX 4.3 BSD to VAX/VMS took over a month, and porting to |the Sun workstation took a week. | |AIX seems to be System V with a few BSD enhancements. The only two |utilities that were missing were "more" and "head". I am told that the |latter can be simulated with "sed" quite easily, and that the former is |replaced by "pg". Nonetheless, I wrote my own "more" and "head" anyway. | |Disclaimer: The machine and AIX are on loan from IBM, but this report | is strictly my own opinion. |From: ames!cs.utexas.edu!uunet.UU.NET!mcvax!rhi.hi.is!frisk (Fridrik Skulason) |Organization: University of Iceland (RHI) | |We have been using AIX a bit here at the University, but only on one type |of machine (PS2/70). We have been comparing AIX to other systems like |INTERACTIVE/Unix, and found AIX to be our system of choice. | |Graphics - Well, we are running X/Windows on it right now, with no (serious) |problems. | |I am not able to give you much more information, but you could try contacting |the person in charge of the testing (goggi@rhi.hi.is) |From: Sherri Menees Nichols <sbcs!smn+@andrew.cmu.edu> | |I have AIX running on a PS/2 Model 70, and it's pretty reasonable. I'm |not sure how far down the PS/2 line AIX runs, though I read nothing in |the documentation to suggest that it didn't run on all PS/2's. AIX is |also available on the RT, but it's a different implementation, done by a |completely independant group. My understanding is that IBM wants one |implementation of AIX that runs on everything from PS/2's to 370's, but |it's going to be a while before that happens. | |I've not used System V, only BSD. AIX is easier to install than BSD, |though still tricky in places. As far as performance, well, my machine |has a 120 meg disk and 8 megs of RAM, so it's tough to find anything |that doesn't perform well. I have an 8514 display with a VGA adapter, |and there's supposedly a higher-resolution adapter on the way. I'm |running X-windows. I also have DOS Merge, which allows you to run DOS |processes on top of AIX, sharing the AIX file system. Pretty |convenient. |From: Andrew M. Winkler <sbcs!amw@cunixb.cc.columbia.edu> | |If your only choices are AIX and VM, it wouldn't really matter _how_ bad AIX |might be... I have used it on RT's at Yorktown, and it's not too bad. It's not |BSD, but the rumors suggest it will become more so. (IBM has a BSD, but only |sells it to universities.) |From: ibmarc!ks!drake@sd2.almaden.ibm.com (Sam Drake) |Organization: IBM Almaden Research Center, San Jose | |I like AIX, personally. | |AIX is currently available on the IBM RT (RISC machine) and on the IBM PS/2 |models 70 and 80 (models with Intel 386 processors). Versions for the 386SX |and 486 versions of the PS/2 have been announced but are not yet shipping, |and a version for the IBM 370 family (mainframes) will be available later |this year. | |AIX is a reasonable mix of System V.2 and BSD; I can take BSD sources off |of usenet and get them going pretty quickly. | |IBM has a fully supported X Window System Version 11.3 for both the PS/2 and |the RT that works pretty well. The RT supports other graphics standards |as well, such as graPHIGS. | |I'm not a salesman, nor an official spokesman of IBM; hope the basic facts |above help you get your investigation off the ground. |From: rifrank@yktvmh (Ronald I. Frank) | T.J. Watson Research Center H3-B40 | Box 704 | Yorktown Heights, NY 10598 | |I am in a small group (about 12 people). We, and 2-300 other AIX (tm) users |depend on a service group of 6 gurus for help. (We are part of the |Computer Science Dept. of the IBM T.J. Watson Research Center which is |located at Hawthorne). I am a Mathematician/Computer Scientist who has |recently switched into Virtual Environment research work using |workstations and parallel computers. Being a small group we do much of |our own grunt work - cf. below. | |I am not a guru, being only on AIX for about 7 months. However, I have |been in computing for close to 32 years using many machines, operating |environments, development environments, and languages. I also teach |Software Engineering in a local graduate department - based on both |large system and PC experience. | |Reply: | |Note my address. Given that, any IBM branch office can give you details |but here is an outline from a user. I am sitting at an RT running AIX |rel 2.2 with X-11 (.3?). I am composing this on an IBM 3090 running |VM/XA - only because I'm new to UNIX (tm). and I still prefer my local and |powerful host tools. The 3090 host is my internet node via Bitnet and |CSNET. I could use my AIX RT as my node by I have not had to - so I don't. |I'm rifrank@ibm.com . Some of my colleagues have switched over to using |only their workstation as their node. | |I have 6 shells running including emacs, an rlogin to an ARDENT graphics |supermini that I do some admin on for my group, plus a few background |processes running some local email and service functions. I see standard |X-11 menus, monitors, and clocks (I like both the analog and digital). |I use a color 5081 19" megapel tube - and csh. My emacs is GNU emacs. |I also use vi. | |AIX 2.2 is V.2. It has BSD features also. We have a pair of AIX RTs on |the ARDENT, They do data acquisition in real-time for gesture capture. |They use a thin ethernet and sockets for comm. to the ARDENT. Both |interface using RS-232-C, one to VPL gloves and one to a Polhemus 3-D |tracker. The idea is to off-load the ARDENT which couldn't handle both |the data gathering and the real time graphics for visualization. | |Another AIX-RT is a router in our ARDENT lab, on the first floor, |connecting to a building IBM token lan. The ARDENT is also serviced by |an AIX RT running as an NFS file server. We have /u (our users) out on |the file server. All of the development code is shared by getting to |the server via the router and the net. All these local RTs are connected |to the ARDENT by the cheapernet, the router is also connected to the |token lan on its outside. I am connected by an IBM token ring to a floor |router on the 3rd floor. I ftp to and from my 3090 or the ARDENT (it |runs V.3). I rlogin to the ARDENT, the RT file server, the RT router, |the RTs with the gloves and the tracker, and my colleagues' machines - |if needed. I do my admin chores in V.3 on the ARDENT and my RT chores |in V.2 on the RT. | |We have a few Suns and other workstations around here on the same lan |structure. We got them before AIX RT was available. AIX seems the |same as the other system V derived UNICES to the user. I have recently |installed AIX on a PS/2 80 (a 386 box) I took home. It looks and feels |the same - only the megapel display I have at the office is crisper |(higher res) than the 8514 (3/4 Mpel) I have at home. We also have AIX |running on one of our 3090s. I haven't used it yet but I hear it is |pretty much the same. AIX PS/2 is a little more Berkeleyish. If |it means anything, I use the IBM doc for my AIXs. I have the AT&T and |the Berkeley books, which I sometimes reference. I Like the O'Reilly |nutshell guides. The point is - AIX is a standard UNIX V - with |improvements. | |I keep my RT up 24 hrs a day since we have a local backup that |cron runs at night to the host. I don't back up to tape since the |host guys back up to mass tape every night. (Last night when I |started this note we had a storm and a total power failure. All the |RTs and AIXs came back without a problem.) | |I print on an IBM 3812 page printer (lan connected) outside my door. |Some people use TEX. I don't - yet. I do my sys admin with the ARDENT |folk over the internet, sending them bug examples and getting back fixes. | |The RT does very nice graphics; we have the mandril, fruit bowl, and |other -unmentionable - standard graphics demos; however for massive |real-time graphics ("visualization" work) we use IRISs, ARDENTS, PIXARS, |etc. - all lan connected. | |I miss the powerful tools of the host and of PC-land but UNIX is slowly |catching up. I am primarily a research computer scientist. I code now |in C on the RT. We have the usual tools (make, sccs, etc). We also |have some very powerful internal tools that I hope make it out someday. |The C compiler is the standard UNIX C. | |I am starting to do some work in C++. I have the Oasys C++ on my RT and |the UNIPRESS CMACS. I am not necessarily recommending any of these |products since I have not yet had time to use them much. We also have a |home grown C++ port for the RT. I don't think its available outside. |The guy who did the port says it took him a couple of days - part time. |We don't have C++ on the ARDENT. I just got a source license for it from |AT&T so that we can port it to the ARDENT. We are using the AIX on the |RTs to do the port (also the ARDENT of course). | |PROMO INFO (things I haven't yet used much). | |AIX has a (PC)DOS capability to run multiple DOS sessions, each using |full memory - concurrently. PS/2 and host AIX also has the LOCUS Corp. |capability on the host (transparent file system across the net). See |your local rep for the details. | |I hope this has been of some use to you. I have only the V.3 UNIX on the |ARDENT to compare with, but it and my AIXs (RT & PS/2) are essentially |similar from my user's perspective. |From: Vince Guarna <iuvax!uiucuxc!uicsrd.csrd.uiuc.edu!guarna> |University of Illinois, Center for Supercomputing Research and Development | |I have been using AIX for a couple of weeks now. I am running on a PS/2 |Model 80, 4 Meg RAM, 115 Meg disk. Running C, Fortran, X Windows, DosMerge, |and the usual other stuff. I think the news is mostly good. I have had |relatively good success porting software to it, both public domain unix-style |things and X Windows DOS things. I have no direct experience with |System V.3, but for a year, I was running Xenix 2.2. | |The AIX C compiler is significantly better than Xenix's. Not talking |performance here, just the parser. Xenix has several problems in taking |standard C constructs (at least in version 2.2). | |System performance is OK, but if you run an extremely disk-intensive |application (such as tarring/untarring to/from floppies), forget about |anything else -- even keypress echoing. | |Xenix has a "multiscreen" feature: Alt-Fn takes you to tty n. This is |significantly better than AIX's multiscreens...here you must cycle through |each screen in sequence to get to the tty you want. | |Vi performance in AIX is slow for some reason. Leaving insert mode with |the ESC key takes a half second or more. | |CSH is infinitely better on AIX. True pushd/popd, job control, of course, |stty crt works (real Control-U). | |Several commands missing on AIX: more, strings, lpr, man. | |nroff comes only with man and mm macros -- no ms or me. At least Xenix |came with ms. | |AIX system administration is very wacky -- they tried to make things |easier for the non-unix type, I guess. In the process, they made it much |tougher for those familiar with the more conventional system admin tools. | |Serial port operation much better on Xenix than AIX. In Xenix, you can |have a single serial line that is enabled for modem dialins and still |use the line for dialouts without reconfiguring the device. I have |not been able to achieve this yet with AIX (could be my ignorance..dunno yet). | |The IBM C compiler puts its damn 5-line copyright message out everytime |the parser is called. This is a major nuisance when you build applications |consisting of 300 files and want to keep a logfile from the make. | |The "devices" command (configuring devices ... part of their "unique" |system admin) is inadequately documented. You are prompted to input |characteristics that are not explained anywhere. | |X windows is OK -- no major porting problems yet. Performance seems better |on some things like window move/resize than it does on a Sun 3/50. However, |some things like popup menus are much slower. It is difficult to say |how much of the performance problems are caused by insufficient memory. |I am running 4 Meg, and probably need to go to 8. | |DOS Merge works pretty well -- I have found a couple of applications that |will not work (like kermit! -- causes a "divide exception"), and some |that degrade in performance by one or two orders of magnitude (Tex |previewer), but for the most part, it is pretty slick. The merged |filesystem/environment is excellent. To be fair, I did not try |VP/IX from SCO...might have been just as good. | |The system is huge. By the time I installed all the software, set up |the swap and /tmp space, etc, I had about 36 Meg left for me (remember |that I started with 115). And this does not include man pages. This is |not enough for the kind of work that I do, so I will probably be buying |another disk. For a turnkey application, tho, you could probably get |away with a 60 Meg disk. | |The UUCP is the Honey-Danber variety. I used the conventional before, |but this seems just as good if not better. | |LOTS of good Berkeley stuff in AIX -- long file names, soft links, already |mentioned job control, sockets (I tried these out, they seem to work). |They still use a lot of System V things, tho, like flags on "ps", stdio.h. | |My overall opinion: I am glad I made the switch. Long filenames are actually |pretty important for me -- especially since I use SCCS which takes away |two chars from the 14 you have in SYSV. The serial line is the only |headache, and I think it is a very major headache. Having to disable the |tty line to let my UUCP call out is not good. I hope that I will be able |to get rid of this problem with a few hacks/scripts or whatever over the |next few weeks. |From: Andrew Dingwall <andrew@root.co.uk> | |I believe AIX runs (or will run one day) on machines from the RT up to |quite large machines (3090 etc). | |Here are a few differences that I observed while porting some software |to an RT (IBM 6150) running AIX version 2.1.2: | |AIX comes 'unbundled'; ie: you have to buy several packages to get a |reasonable system. Even vi(1) is not in the basic distribution, but comes as |part of one of the add-on bits. | |Not many of the common Berkley-derived utilities are supplied |(eg: more(1) and strings(1) are not supplied) | |There seems to be some general braindamage relating to the differences |between upper- and lower-case characters. sort(1) behaves as if the -f option |is in effect all the time, ls(1) treats cases as equal when listing files, |commands in the grep family have lost the -i option. | |Signal handling seems to be SysV with some BSD extensions; ie: most signals |are reset when caught but sigblock(2), sigvec(2) etc are provided as well. |Some of the signals have different names. | |ftok(3) complains when the second argument is 0. This is documented but is |not standard. | |I found that it is quite easy to run out of temporary file space on a |standard configuration machine when compiling largish programs. |This is because the C compiler generates huge temp files. |It is not possible to override this as in V.3 by redefining TMPDIR. |The optimiser will sometimes loop on programs containing long switch |statements where a reasonable number of case values are randomly distributed |over (say) 1 to 10,000. | |I did not seem to be able to get adb(1) to work as one would expect it to. | |I was unable to get IBM to tell me how to drive the system console using |escape sequences (a la termcap/terminfo) - instead they tried to get me to |use some fancy (AIX-specific) library routines. (However, a bit of informed |guesswork sufficed here!) The termcap file is not supplied. |Some terminfo files are supplied but only for a few IBM terminals, |DEC vt100/220 series and the tvi925. Terminfo source is not supplied. | |On the RT console keyboard, not all the keys generate the characters that |you might expect. In particular, '#' (hash) and '|' (Unix pipe) |echo correctly on the screen but do not generate the correct ASCII code. |This can probably be changed by some setup routine but IBM technical |support were unable to tell us how to do it. | |Although BSD-style sockets, TCP/IP and ethernet appear to be supported, |the combined conclusion of our networking gurus and IBM technical support |seemed to be that AIX systems are not intended to talk sensibly to the |rest of the Unix network world. However, there would appear to be quite |a lot of support for communication with other AIX machines (and even other |IBM non-AIX machines) using SNA and LU protocols (yes, even LU6.2). | |The documentation is extensive (by Unix standards) but its layout departs |somewhat from what Unix people are used to - this makes things hard to find. |Also, we found that the manuals didn't wear very well; pages tend to fall |out rather easily. | |System admin is a pain - there are lots of funnies here. | |Conclusion: expect a large learning curve and don't assume too much (as if |you thought you knew Unix! :-)) |From: ames!rutgers!cse.ogc.edu!servio!penneyj (D. Jason Penney) |Organization: Servio Logic Development Corp.; Beaverton, OR | |We have ported our GemStone software from VAX/VMS to SunOS and more |recently, AIX for the IBM RT. The RT's debugging tools are rather lame. |Since the RT doesn't have virtual memory, wild memory stores can corrupt |the kernel. If we didn't have the other hosts to debug our software, |RT development would be impossible. | |The RT is also rather slow. I think the Model 80 would be better, but |we haven't played with it. | |STANDARD DISCLAIMER: Should I or my opinions be caught or killed, the |company will disavow any knowledge of my actions...
dyer@spdcc.COM (Steve Dyer) (07/23/89)
One thing that wasn't clear from people's comments is that AIX PS/2,
a Locus kernel with Sys V.2 stuff grafted in, is pretty much a
different animal than AIX RT. It also has a fair number of features
that aren't shared by AIX RT; right off the top of my head, I'd
enumerate job control and >14 character file names on the native
file system. It's not very useful to confuse the two of them.
--
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer@arktouros.mit.edupapa@pollux.usc.edu (Marco Papa) (07/23/89)
In article <3829@ursa-major.SPDCC.COM> dyer@ursa-major.spdcc.COM (Steve Dyer) writes: >One thing that wasn't clear from people's comments is that AIX PS/2, >a Locus kernel with Sys V.2 stuff grafted in, is pretty much a >different animal than AIX RT. It also has a fair number of features >that aren't shared by AIX RT; right off the top of my head, I'd >enumerate job control and >14 character file names on the native >file system. It's not very useful to confuse the two of them. Very true. AIX RT is System V based, AIX PS/2 was built starting from a 4.3-like kernel, and as a result is much more compatible with 4.3 than AIX RT. While porting magic to AIX RT required lots of changes, almost nothing was needed to port it to AIX PS/2. -- Marco Papa USC- ACSC -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= uucp:...!pollux!papa BIX:papa ARPAnet:pollux!papa@oberon.usc.edu "There's Alpha, Beta, Gamma, Diga and Caligari!" -- Rick Unland -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
dd@beta.lanl.gov (Dan Davison) (07/29/89)
The only thing you need to know about AIX is that they swap the *kernel*. Can you say geologic time to run things? I knew you could. -- dan davison/theoretical biology/t-10 ms k710/los alamos national laboratory los alamos, nm 87545/dd@lanl.gov (arpa)/dd@lanl.uucp(new)/..cmcl2!lanl!dd
drake@ibmarc.uucp (Sam Drake/99999999) (07/30/89)
In article <29176@beta.lanl.gov> dd@beta.lanl.gov (Dan Davison) writes: >The only thing you need to know about AIX is that they swap the >*kernel*. Can you say geologic time to run things? I knew you >could. As an IBMer (but not an IBM spokesman), perhaps I shouldn't respond to this, but here goes anyway. The implied statements here are that (1) AIX/RT is slow, and that (2) it's slow because the kernel is swapped. I don't believe that either one of these statements is especially accurate. I won't use this append for sales purposes, but will simply comment that AIX/RT systems have been benchmarked in the UNIX magazines, and the numbers look fine to me...read 'em yourself and judge. It is certainly true that much of the AIX/RT kernel is pagable. The lower level VRM interface, on which the AIX system resides, is permanently resident and never paged. Claiming that this leads to a performance problem is not accurate, in my opinion. Why should all pages of the kernel be permanently resident in storage? If a particular page of the kernel is the Least Recently Used page in the system, why leave it in storage? To leave it in, though it's not been referenced for hours, means that system performance will suffer, since pages that have been recently used are paged out to disk instead of this more deserving page. Is it really important that the text of the kernel panic handler be kept around, hogging up real memory, when in reality it will be referenced far less than once per MONTH? I don't think so. Granted, if a kernel panic occurs, one more disk I/O will be required before the system abend dump can be taken, but I think that's a fair tradeoff for having that page available to user processes for all those thousands of seconds. If main memory is so badly over-committed that useful pieces of the kernel are being paged, then the situation is of course not as obvious. But even in the case of such over-commitment the principle of paging the Least Recently Used pages first is a winner. If it's been 20 seconds since the serial port device driver has been used, get it out of there! Use the memory for running more vital things (like the user's application). Many if not most operating systems have a resident portion and a pageable portion. All do it because LRU beats non-pageable kernels. A non pageable kernel is certainly easier to write, but it's not a performance boon ... quite the opposite. Cheers, ...Sam Drake / IBM Almaden Research Center Sam Drake / IBM Almaden Research Center
gors@well.UUCP (Gordon Stewart) (07/31/89)
I have found no performance problems with the RT under AIX -- it's
certainly faster than a Sun 3/260, etc. I have plenty of complaints
about AIX, however --
I don't particularly like the documentation -- especially the user
docs, which give a syntax diagram of the commands, but are sparse
in examples (usually, they'll give one or two trivial examples). When
my colleagues and I asked IBM if they'd ever seen Sun's documentation
(which is EXEMPLARY) or AT&T's (ADEQUATE) they responded by claiming
"our documentation is the best in the industry"
I especially don't like that they've changed the name of some common
commands (if you use AIX, you know which ones) and added multi-letter
command-line options -- a clear poke in the eye to System V, which
actually has STANDARDS about commands and options. I guess IBM felt
the need to put their proprietary stamp on it -- like what Raymond
Chandler said about editors (they like the flavor better after they've
pissed in it)!
Product Support and Defect Support are abysmal -- at least from the
point of view of a competent UNIX programmer trying to develop
software for the RT under AIX. I must go through too many hoops, and
speak to too many people who don't know diddly-poop about what I'm
saying, before I get a response from someone who's informed. The
usual IBM chain of command B.S.
Typical of IBM, they have the most polished sales and marketing
presentations in the business -- targeted for the person making
purchase decisions. The rest of us have to live with those decisions,
however.
All in all, the RT is a pretty good piece of hardware -- but AIX isn't
a mature enough product to take seriously. And the "improvements" and
"enhancements" that make AIX proprietary (i.e., non-standard!) will
have the effect of relegating AIX to a very small percentage of the
UNIX market share. Which means less support, any way you look at it.
Just look at the traffic in this newsgroup, compared to comp.unix.whatever!
Not all articles are of note, but I have found a few here that were
very helpful, and am grateful to those who have responded to my cries
for help. There are just too damn few of us, though!
A SUN4! A SUN4! My KINGDOM FOR A SUN4!
Michael Sierchio
--
{apple, pacbell, hplabs, ucbvax}!well!gors
gors@well.sf.ca.us
(Doolan) | (Meyer) | (Sierchio) | (Stewart)drake@ibmarc.uucp (Sam Drake/99999999) (08/02/89)
In article <12938@well.UUCP> gors@well.UUCP (Gordon Stewart) writes: > >I especially don't like that they've changed the name of some common >commands (if you use AIX, you know which ones) and added multi-letter >command-line options -- a clear poke in the eye to System V, which >actually has STANDARDS about commands and options. Can someone give me an example or two of each of these? What command names are non-standard in AIX/RT 2.2.1? What commands have multi-letter command line options? Truly curious, ...Sam Drake / IBM Almaden Research Center / drake@ibm.com Sam Drake / IBM Almaden Research Center
fleming@balboa (Dennis Paul Fleming) (08/05/89)
In article <977@ks.UUCP> drake@ibmarc.UUCP (Sam Drake) writes: >In article <12938@well.UUCP> gors@well.UUCP (Gordon Stewart) writes: >> >>I especially don't like that they've changed the name of some common >>commands (if you use AIX, you know which ones) and added multi-letter >>command-line options -- a clear poke in the eye to System V, which >>actually has STANDARDS about commands and options. > >Can someone give me an example or two of each of these? What command >names are non-standard in AIX/RT 2.2.1? What commands have multi-letter >command line options? The two most frustrating in my experience where telnet (called tn in AIX) and ftp (given the name xftp) Why IBM would shorten one command name and lengthen the other is beyond me. the print command has several multi-letter command line options: -bp, -ca, etc.
drake@ibmarc.uucp (Sam Drake) (08/09/89)
In article <2461@orion.cf.uci.edu> fleming@balboa.UUCP (Dennis Paul Fleming) writes: > The two most frustrating in my experience where > telnet (called tn in AIX) and > ftp (given the name xftp) > > Why IBM would shorten one command name and lengthen the other >is beyond me. Please consider upgrading your system to the current release of AIX/RT. Earlier releases had that problem, but the current release uses the BSD names entirely. Sam Drake / IBM Almaden Research Center