Hi Charles, I was just responding to your basically unilateral statement that "I can confirm that with typical SBO usage Microsoft SMB/CIFS has terrible performance." (I'm not questioning your direct experience, by the way; it would be interesting though to know when you did the testing?) That may have been the case with SMB in the past, but I know things have drastically improved from the SMB1/SMB2 days... And the OP did say he was looking at using Win Svr 2012R2 (which implements SMB 3.02.)
I too follow Foskett (more on the networking side of what he puts on with TFD than with the storage side), and I don't think he's astroturfing for MSFT... I just think he was really impressed (perhaps with pre-release data, but I haven't see anything from him since that counters this...) It just makes me crazy when people say that MSFT tech sucks today because of their experience with it back in the NT4.0/Win 2000 Server days, and assume it's still the same way today... Technologically, they've improved their tech a whole bunch since the NT4/2K days (Windows Server, Active Directory, SQL Server, Exchange Server, scripting languages, you name it...) (and by the way, I love, use, and appreciate *nix too; I'll use whatever the business requires to meet its goals.) Best, Will -----Original Message----- From: Charles Polisher [mailto:cpol...@surewest.net] Sent: Friday, May 23, 2014 2:13 AM To: Will Dennis Cc: Tim Kirby; LOPSA Tech Subject: Re: [lopsa-tech] MS Windows as NFS Server? Will Dennis wrote: > Which version of SMB are we talking about here? (CIFS == pre SMB 1.0, > i.e. the NT4.0 proto) MSFT is shipping SMB 3.02 now on Server 2012 R2, > which is extremely high-performance (so much so that they allow SQL > Server DB file access as well as Hyper-V VM file access over SMB 3.x > as an alternative to iSCSI or FC > connections...) Great question. (BTW, SMB and CIFS are often used interchangably by Microsoft. Their definitive protocol specs (MS-CIFS, MS-SMB, and MS-SMB2) sanction interchangable usage. I haven't read MS-SMB3.) > Please see > http://blogs.technet.com/b/josebda/archive/2013/10/02/windows-server-2 > 012-r2-which-version-of-the-smb-protocol-smb-1-0-smb-2-0-smb-2-1-smb-3 > -0-or-smb-3-02-you-are-using.aspx for further details on versions and > history. > Please also take a look at > http://blog.fosketts.net/2012/05/06/smb-3-huge-scope-impact/ for > Stephen Foskett's take on SMB 3.0 (Stephen is a storage expert, > independent blogger, and runs Storage Field Day as a part of his Tech > Field Day series...) > Facts, not FUD, please. I'm merely reporting my own experience on a large heterogenous net with actual servers and clients using these protocols in production. In the right environment with the right clients and servers, SMB 3 probably rocks. However, I was responding to a comment (which you trimmed) about CIFS performance in general, which I expanded on. The Foskett article cited (I happen to be a Foskett fan) has a lot of superlatives, was written prior to the product release, and appears to be based entirely on product claims -- not observed performance. He could be right, but I wouldn't call my observations FUD, especially not while referencing Foskett's hype! > (note that the above does not speak to the OP's original q on NFS on > Win2012R2, and suggestions to consider SMB instead - just trying to > make the point that SMB is a performant proto vs. NFS these days...) _______________________________________________ Tech mailing list Tech@lists.lopsa.org https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
