Interoperability in the real world

gallery

Too many users save their documents in binary formats that are both proprietary and transitory.

The justification for this practice is that the proprietary formats are 'de facto' standards. "De facto they may be," according to Jeremy Allison, a lead developer on the Samba team, "but standards they are not." The Samba server allows Linux and Unix to serve and share files on Windows networks, using Microsoft's SMB/CIFS (Server Message Block/Common Internet File System) protocol. SMB/CIFS is Microsoft's standard implementation that allows communication on PC networks running Microsoft systems. Linux and Samba drive most of the dedicated print and file servers on the market.

SMB/CIFS is described by Allison as "a bizarre hybrid between an open and a proprietary standard." The implementation on Windows deviates widely from the published standard. "Microsoft engineers tend to see the protocol as unique to their own systems", says Allison. "They view the implementation on Windows as correct even where it deviates from the X/Open standard." The Samba team interprets these deviations as bugs, and Allison describes Samba as "bug for bug compatible" with Microsoft. "The spec was a fiction, and made no sense. The complexity of SMB is one of the main reasons that NT was so unstable."

Although SMB/CIFS is a set of networking protocols for exchanging data between computers, the Samba team had the choice of signing an NDA, or reverse engineering the Windows implementation. An NDA was out of the question because Samba is free software and the code is visible. "Samba can exist because Microsoft has to provide backwards compatibility with all their older systems," Allison notes. "We live in their backwards compatibility shadow, which is very long indeed. We can emulate those file-sharing protocols because they can't change the clients that much and break the backend servers," but this solution is far from satisfactory. In laboratory tests, conducted by IT Week labs in October 2003, Samba on Red Hat Linux outperformed Windows Server 2003 as a Windows domain controller by a factor of 2.5, but the success of Samba does not disguise the fact that the absence of an open standard impedes interoperability.

"The SMB protocol is disgusting", says Allison. "It grew like a wart. You can tell. It has a 39-byte header, which might have been important when bytes were important on the wire, but now it's crazy. Eventually, I'd like Samba to go away."

I can't talk your language

Proprietary data formats, published or not, offer little long term data security, and are an unreliable choice for transferring data to prospective clients. A few years from now, the contents of the current word processor document, for instance, may well be hidden from view, and will be messy and expensive to retrieve. Reliance is too often placed on backups and short term solutions, and inevitably the data will be lost and forgotten - and the search will be frustrated by the inability to read the binary format that was the "de facto standard" a few short years before.

Microsoft Office, for instance, currently dominates large sectors of the market for productivity suites, possessing the lion's share of the customer base, but this has only been true for the last ten years or so. Back in the '80s, Lotus 1-2-3 had over 90 per cent of the spreadsheet market and Excel was a clunky no-hoper. WordPerfect was a billion dollar giant, and Word had less than 10 per cent of the market. The subsequent success of Word and Excel probably owes more to the success of DOS and Windows than it does to the inherent virtues of the products themselves.

Email to a friend

Print this page