Samba

So many Samba successes

In 2024, the Samba Project is nearly 30 years old and has conservatively a billion users. It started by discovering some unmaintained but interesting open source software for sharing files and printers with workstation computers. Microsoft wanted to monopolise all computer networking and I didn't know what else to do! This history is documented in the official Samba repository.

Samba wasn't just a clever idea, it was implemented by some very talented computer scientists and software engineers with a very large number of total contributors. I was not one of the Samba core engineers, not being good at intensive protocol analysis combining encryption, obfuscation, historical anomalies and sheer overwhelming volume of old-school RPC design. I was (and remain) more interested in interoperability architecture and design, the why these things are needed and make sense to users. For its first decade Samba was one of the highest-prestige projects in the present open-source world is Samba, as described in the influential essay "The Cathedral and the Bazaar". Certainly Samba was a factor in holding back absolute hegemony by Microsoft, despite their aggressive attempts to destroy compatible technologies.

Samba became a story of adversarial interoperability, protocol analysis often called reverse engineering, IP rights, threats from Microsoft, cybersecurity, a giant European court case, startup companies, lawyers and engineering excellence. I got involved with Samba because I needed it to solve my own problem sharing files and printers at the University of South Australia, which was increasingly having its infrastructure taken over by the Microsoft monopoly. I could see there was a bright future for drop-in replacements for Microsoft network servers, and fortunately some engineers with talents beyond my own agreed with me. It was an interesting ride for many years!

Samba was the first software to have the right of compatibility affirmed by the EU Court of Justice, after an epic series of cases finishing in 2012. The EU Commission learned then that it could fight a giant American tech corporation and win, something it is continuing today in battles around monopolies, privacy and dumping.

Comprehensive failure by its own measure

Samba has fallen far short of its promise to be a "drop-in replacement" for Microsoft servers in the full sense including ease of deployment, which would have made it ubiquitous in every company and home in the world. Microsoft and Amazon's hybrid cloud solutions would have looked very different. Unfortunately, no matter how advanced Samba becomes now, the opportunity for Samba to rule the world seems to have passed (or has it? see further down for news in 2024 ...)

The Samba Project is thus a thriving open source project with a billion-plus users, which nevertheless fails in its original primary objective.

Samba is still developed and is still impressive. The estimate of a billion users is due to its inclusion in many embedded devices (eg printers, photocopiers and cameras) and giant file storage systems. Only Samba and Microsoft completely implement Microsoft's vastly complex Active Directory specification, which is at the heart of a majority of the world's corporate IT infrastructure. Samba Team engineers continue to release reliable code, with a core team of around 30 members, who are volunteers and also developers funded by many companies.

Reverse and Forwards Engineering/Protocol Analysis

Samba started as a protocol analysis project (not strictly "reverse engineering", which is incorrect as noted above) to provide users with the same experience as having a Microsoft server. There were some additional benefits for users due to being based on Linux and open source. After the astonishing 13-0 loss in the EU Court of First Instance (now the General Court) in 2007, Microsoft started sharing the written standards for how to communicate with their servers. At last we could really know what we were doing in developing Samba! Microsoft would no longer control all file servers and directory servers in the world!

But it didn't quite turn out that way, really disappointingly so.

Part of the reason Samba fell short stems from the social and psychological difficulty of turning a reverse engineering project into forwards engineering. That is inevitably what happened when the full documentation for the SMB protocols became available. The architectural possibilities are very different if you have the documentation. The development cadence should have changed completely to reflect that. Companies were very keen to participate in this new opportunity, and deployments in the cloud were an obvious next step. None of that happened, perhaps because Samba was no longer the only alternative in the market, and was no longer driven by the brilliance of dogged individual discovery. Additional skills, such as user interfaces, were needed for Samba to become as ubiquitous as Microsoft in the server market, and those skills were never applied.

Nevertheless I am very proud of Samba, and I enjoy seeing its continued technical growth. Samba seems like it has plenty of future yet.

Wonderful news in 2024

Germany's Sovereign Tech Fund has has made Samba a substantial grant to be used for protocol engineering. While constant and advanced protocol engineering keeps Samba as a drop-in replacement on the network, or close, this probably still won't make Samba the seamless drop-in product replacement because that would require a user interface project and the Samba code is now elderly. But perhaps it will, and in any case it is possible that ways will be found to keep the large and complicated Samba codebase fresh. Samba has been built on a tiny fraction of the budget of the Microsoft network engineering team even if you count the volunteer hours, and maybe this will balance things out a little. I hope so!