You wrote: I have been disappointed that I can't use Linux+Samba as an alternative to a Win9x or NT server for our GP medical practice. The software we need runs in win9x. We crave stability. But Linux+Samba is just much slower as a file server. It's normally quite fast: are you running over a dial-up network, requiring you to adust MTU's for speed??? If not, try ``socket options = TCP_NODELAY'', which adapts server TCPs to a MS feature (:-)) and produces considerable speed increases... --dave -- David Collier-Brown, | Always do right. This will gratify some people 185 Ellerslie Ave., | and astonish the rest. -- Mark Twain Willowdale, Ontario | davecb@hobbes.ss.org, canada.sun.com N2M 1Y3. 416-223-8968 | http://java.science.yorku.ca/~davecb
>I have been disappointed that I can't use Linux+Samba as an >alternative to a Win9x or NT server for our GP medical practice. >The software we need runs in win9x. >We crave stability. But Linux+Samba is just much slower as a file >server.This is not my experience at all. We are using Infomagic's Workgroup Server, which is not the latest and greatest (based on RedHat 4.1), but we serve up 60 GB to both PC's and Mac's at the same time, with good stability and speed. I had one Linux server running this that I could also boot in NT mode, for backwards compatibility with archive tapes. The same hardware gave me better performance running Linux than NT. InfoMagic claims the WGS is thoroughly debugged, and that seems to be the case. It also has good gui admin tools. The only significant tweak we made was to increase the maximum number of inodes and files, with a script run from rc.local to echo 4096 > /proc/sys/kernel/file-max echo 12288 > /proc/sys/kernel/inode-max You might want to check it out at http://www.infomagic.com . Now one thing I should mention, we do have one NT server in-house, and it is the acting domain controller for the people logging in from PC's. [Samba can't be the domain controller yet, can it?] //\ --rich // \ // / \ Richard Mann -- System Administrator // / / \ P u b l i c a t i o n S e r v i c e s, I n c. \\ / / / Champaign, Illinois, USA 61821 (217) 398-2060 x36 \\ / / \\ / http://www.pubserv.com http://www.primesynergy.com \\/
Hello,>I have been disappointed that I can't use Linux+Samba as an >alternative to a Win9x or NT server for our GP medical practice....>Is it possible that Win98 does TCP/IP properly and might be faster >than win95 as a tcp/ip workstation client??im my experience Win98 isn't any faster than Win95 and sometimes is even slower. Also in my experience correctly configured Samba Servers working with Win95/98 aren't any slower than NT Servers as long as TCP/IP is used. Samba is only a little slower than Novell. Hope it helps. Dr. Sbragion Denis InfoTecna Tel, Fax: +39 039 2324054 URL: http://space.tin.it/internet/dsbragio
> We crave stability. But Linux+Samba is just much slower as a file > server. > > It's normally quite fast: are you running over a > dial-up network, requiring you to adust MTU's for > speed??? > > If not, try ``socket options = TCP_NODELAY'', which > adapts server TCPs to a MS feature (:-)) and produces considerable > speed increases...There is another point I've reported to samba-bugs a few days ago: We've experienced a great *loss* in performance when migrating an Access application (MDB file > 16 MB) from a slow NT-Server (486DX-66, NT 3.51 server) to a muxh faster Samba server (P166, 64 MB, Linux 2.0.35, Samba 1.9.18p10.) The workstations are connected via a 100 MBit fast ethernet. With one workstation using the database it ran normal. With two warkstations it slowed down. With four workstations the application was unusable (> 2 minutes for one query) Playing with smb.conf we've tracked ist down to be a locking problem. Fith 'fake oplocks = yes' it ran at normal speed. (Of cause we've got database consistency problems.) A friend of me looked over the Samba code and told me to reduce the LOCK_RETRY_TIMEOUT in local.h from 100 ms to 5 or 10 ms. After that the application ran at full speed again. The server load is still normal. Maybe 100 ms for the retry timeout are to much wasted time for a fast network... I'm not sure if this was the right way to solve the problem. But it works :-) [We've also set -O6 -m486 -fomit-frame-pointer in the Makefile. Maybe it makes Samba a little bis faster :-)] Ciao Marcus * We build our computers the way we build our cities -- over * time, without a plan, on top of ruins. * (Ellen Ullman, "The dumbing-down of programming")
"Marcus Graf" <mg@graf.weinheim.de> wrote:> There is another point I've reported to samba-bugs a few days ago: > > We've experienced a great *loss* in performance when migrating an > Access application (MDB file > 16 MB) from a slow NT-Server > (486DX-66, NT 3.51 server) to a muxh faster Samba server (P166, 64 > MB, Linux 2.0.35, Samba 1.9.18p10.) The workstations are connected > via a 100 MBit fast ethernet. > > With one workstation using the database it ran normal. With two > warkstations it slowed down. With four workstations the application > was unusable (> 2 minutes for one query)AHA! It would have been easier if you'd mentioned a bit more about the nature of the problem earlier (:-)) You've hit the oplocks issue: you want ``oplocks=no'' on the share with the database, if that's all that's in the share, or ``veto oplock files = this.dbx/that.dbx/theother'' if the ``database'' is individual files in a share. Oplocks give superior speed on non-shared files, by caching them locally. One should never share the files of a database across a network: one should have the clients call the database server instead. If you have a pc database that tries to share across a networked filesystem, you need to turn local caching off, for both performance and consistency. I generally regard single-user file-based PC ``databases'' as being the same sort of thing as free samples of crack cocaine (:-() --dave -- David Collier-Brown, | Always do right. This will gratify some people 185 Ellerslie Ave., | and astonish the rest. -- Mark Twain Willowdale, Ontario | davecb@hobbes.ss.org, canada.sun.com N2M 1Y3. 416-223-8968 | http://java.science.yorku.ca/~davecb
This brings to mind a question though: Why is oplock processing on Samba so much slower than it is on NT? It sounds to me like it should be an architectural problem, but the originator implies that it works reasonably well under NT. Forgive me if this has been addressed previously, but I'm a little new to Samba and SMB networking in general.> -----Original Message----- > From: David Collier-Brown [SMTP:davecb@Canada.Sun.COM] > Sent: Tuesday, October 13, 1998 12:53 PM > To: Multiple recipients of list > Subject: Re: win98 vs win95 with TCP/IP > > "Marcus Graf" <mg@graf.weinheim.de> wrote: > > There is another point I've reported to samba-bugs a few days ago: > > > > We've experienced a great *loss* in performance when migrating an > > Access application (MDB file > 16 MB) from a slow NT-Server > > (486DX-66, NT 3.51 server) to a muxh faster Samba server (P166, 64 > > MB, Linux 2.0.35, Samba 1.9.18p10.) The workstations are connected > > via a 100 MBit fast ethernet. > > > > With one workstation using the database it ran normal. With two > > warkstations it slowed down. With four workstations the application > > was unusable (> 2 minutes for one query) > > AHA! > It would have been easier if you'd mentioned a bit more about > the nature of the problem earlier (:-)) > > You've hit the oplocks issue: you want ``oplocks=no'' on > the share with the database, if that's all that's in > the share, or ``veto oplock files = this.dbx/that.dbx/theother'' > if the ``database'' is individual files in a share. > > Oplocks give superior speed on non-shared files, by > caching them locally. One should never share the > files of a database across a network: one should have the > clients call the database server instead. If you have > a pc database that tries to share across a networked > filesystem, you need to turn local caching off, for > both performance and consistency. > > I generally regard single-user file-based PC ``databases'' as > being the same sort of thing as free samples of crack > cocaine (:-() > > --dave > -- > David Collier-Brown, | Always do right. This will gratify some people > 185 Ellerslie Ave., | and astonish the rest. -- Mark Twain > Willowdale, Ontario | davecb@hobbes.ss.org, canada.sun.com > N2M 1Y3. 416-223-8968 | http://java.science.yorku.ca/~davecb