Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!news.ysu.edu!freenet.akron.oh.us!neoucom.edu!usenet.ins.cwru.edu!pravda.aa.msen.com!nntp.coast.net!howland.reston.ans.net!newsfeed.internetmci.com!btnet!zetnet.co.uk!demon!microl4.microlise.UUCP!news
From: andrewg@microlise.co.uk (Andrew Gierth)
Newsgroups: comp.unix.bsd.netbsd.misc,comp.unix.bsd.bsdi.misc,comp.unix.solaris,comp.unix.aix
Subject: Re: ISP hardware/software choices (performance comparison)
Date: 17 Jan 1996 06:26:18 GMT
Organization: Microlise Engineering Ltd.
Lines: 71
Message-ID: <4di4qa$jf@microl4.microlise.UUCP>
References: <4cmopu$d35@vixen.cso.uiuc.edu> <4cu7t0$mg5@engnews2.Eng.Sun.COM> <4cv8j1$59k@park.uvsc.edu> <4d37d4$j0l@gremlin.backfire.mn.org> <DL29Az.Ax2@ftel.co.uk> <bryDL3r9p.2oq@netcom.com> <4da5nl$8l5@news.voicenet.com> <bryDL5wtq.4Fo@netcom.com> <4dc00a$j2f@park.uvsc.edu> <bryDL7E84.LG3@netcom.com> <4decke$6nh@park.uvsc.edu> <4dg9b0$74d@mail.fwi.uva.nl> <4dh4kc$rnv@park.uvsc.edu> <4di4dc$je@microl4.microlise.UUCP>
Reply-To: andrewg@microlise.co.uk (Andrew Gierth)
NNTP-Posting-Host: andrewg.microlise.uucp
X-NNTP-Posting-Host: microl.demon.co.uk
X-Newsreader: IBM NewsReader/2 v1.2.5
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.netbsd.misc:2007 comp.unix.bsd.bsdi.misc:2161 comp.unix.solaris:57489 comp.unix.aix:68917
In <4dh4kc$rnv@park.uvsc.edu>, Terry Lambert <terry@lambert.org> writes:
>casper@fwi.uva.nl (Casper H.S. Dik) wrote:
>] Terry Lambert <terry@lambert.org> writes:
>]
>] >AIX features that are more modern than equivalents in Solaris:
>]
>] >1) Dynamically sized kernel structures. Solaris has made
>] > some headway in this direction, but AIX beats *everyone*
>] > else cold.
>]
>] Examples please? Can stuff be grown arbitrarily and what stuff can
>] so be grown?
>
>Available vnodes (gnodes) for one. There are other articles based
>on this thread which list others.
Process table has an upper bound of 131072 (on 3.2.5, anyway).
I believe the file table limit is something like 200000. (70000 open
file table entries seems to cause no trouble at all.) I presume the
vnode/gnode tables, record lock tables, etc. have similarly large
limits. All these tables are managed entirely by the system; no
operator intervention is ever required to manipulate them.
The SysV IPC stuff (shared memory, semaphores, message queues) all
have fixed upper bounds, but enormous ones (4096 ids of each type,
65535 semaphores/id, 65535 bytes/msg, 256MB/memory segment, etc.)
The only awkward one is the limit of 10 (?) attached shared memory
segments per process, which is more to do with the architecture than
AIX. However, the limit on mmap() is much higher.
There are so few tunable limits in AIX that it can be quite a shock
to someone used to other Unix variants:
- maximum processes per user (this can be increased without a reboot)
- maximum memory pages available for mbufs
- memory pages assigned to the block-device buffer pool
(NOTE: the filesystem does NOT use these - typically only 20 pages
are assigned to this pool)
The only other tunables are performance-related or administrative.
[And very, VERY few require a reboot to take effect.]
AIX gets away with this because so much of the kernel is pageable
(on systems I've looked at, something like half of the total kernel
memory usage, but will vary).
>] >10) A kernel that is *never* recompiled by the customer.
>]
>] Same with Solaris 2.x.
>
>Point secundus.
I've never used Solaris, but can it match this scenario:
Assuming that hardware can be hot-swapped, it is (I think) possible
with AIX to install an entirely new disk device (for example), for
which drivers were not previously on the system, load the necessary
drivers, configure the new device, migrate logical volumes from
existing disks to the new device, take the old disks offline, remove
their drivers (if appropriate), ALL WITHOUT DISTURBING THE USERS!
[And without requiring any additional software!]
[Anyone done this? I've only dealt with small RS6000s, but all the
evidence is that this is possible]
-- Andrew (andrewg@microlise.co.uk)
[.sig under construction]