Feb 28, 2020
Todays episode is on one of the topics I am probably the most intimate with that we’ll cover: the evolution of the Apple servers and then the rapid pivot towards a much more mobility-focused offering. Early Macs in 1984 shipped with AppleTalk. These could act as a server or workstation. But after a few years, engineers realized that Apple needed a dedicated server platform. Apple has had a server product starting in 1987 that lives on to today. At Ease had some file and print sharing options. But the old AppleShare (later called AppleShare IP server was primarily used to provide network resources to the Mac from 1986 to 2000, with file sharing being the main service offered. There were basically two options. At Ease, which ran on the early Mac operating systems and A/UX, or Apple Unix. This brought paged memory management and could run on the Macintosh II through the Centris Macs. Apple Unix shipped from 1988 to 1995 and had been based on System V. It was a solidly performing TCP/IP machine and introduced the world of POSIX. Apple Unix could emulate Mac apps and once you were under the hood, you could do pretty much anything you might do in another Unix environment. Apple also took a stab at early server hardware in the form of the Apple Network Server, which was announced in 1995 when Apple Unix went away, for the Quadra 950 and a PowerPC server sold from 1996 to 1997, although the name was used all the way until 2003. While these things were much more powerful and came with modern hardware, they didn’t run the Mac OS but ran another Unix type of operating system, AIX, which had begun life at about the same time as Apple Unix and was another System V variant, but had much more work done and given financial issues at Apple and the Taligent relationship between Apple and IBM to build a successor to Mac OS and OS/2, it made sense to work together on the project. Meanwhile, At Ease continued to evolve and Apple eventually shipped a new offering in the form of AppleShare IP, which worked up until 9.2.2. In an era before, as an example, you needed to require SMTP authentication, AppleShare IP was easily used for everything from file sharing services to mail services. An older Quadra made for a great mail server so your company could stop paying an ISP for some weird email address like that AOL address you got in college, and get your own domain in 1999! And if you needed more, you could easily slap some third party software on the hosts, like if you actually wanted SMTP authentication so your server didn’t get used to route this weird thing called spam, you could install Communigator or later Communigate Pro. Keep in mind that many of the engineers from NeXT after Steve Jobs left Apple had remained friends with engineers from Apple. Some still actually work at Apple. Serving services was a central need for NEXTSTEP and OPENSTEP systems. The UNIX underpinnings made it possible to compile a number of open source software packages and the first web server was hosted by Tim Berners Lee on a NeXTcube. During the transition over to Apple, AppleShare IP and services from NeXT were made to look and feel similarly and turned into Rhapsody from around 1999 and then Mac OS X Server from around 2000. The first few releases of Mac OS X Server, represented a learning curve for many classic Apple admins, and in fact caused a generational shift in who administered the systems. John Welch wrote books in 2000 and 2002 that helped administrators get up to speed. The Xserve was released in 2002 and the Xserve RAID was released in 2003. It took time, but a community began to form around these products. The Xserve would go from a G3 to a G4. The late Michael Bartosh compiled a seminal work in “Essential Mac OS X Panther Server Administration” for O’Reilly Media in 2005. I released my first book called The Mac Tiger Server Black Book in 2006. The server was enjoying a huge upswing in use. Schoun Regan and Kevin White wrote a Visual QuickStart for Panther Server. Schoun wrote one for Tiger Server. The platform was growing. People were interested. Small businesses, schools, universities, art departments in bigger companies. The Xserve would go from a G4 to an Intel processor and we would get cluster nodes to offload processing power from more expensive servers. Up until this point, Apple never publicly acknowledged that businesses or enterprises used their device so the rise of the Xserve advertising was the first time we saw that acknowledgement. Apple continued to improve the product with new services up until 2009 with Mac OS X Server 10.6. At this point, Apple included most services necessary for running a standard IT department for small and medium sized business in the product, including web (in the form of Apache), mail, groupware, DHCP, DNS, directory services, file sharing, and even web and wiki services. There were also edge case services such as Podcast Producer for automating video and content workflows, Xsan, a clustered file system, and in 2009 even purchased a company called Artbox, whose product was rebranded as Final Cut Server. Apple now had multiple awesome, stable products. Dozens of books and websites were helping built a community and growing knowledge of the platform. But that was a turning point. Around that same time Apple had been working towards the iPad, released in 2010 (although arguably the Knowledge Navigator was the first iteration, conceptualized in 1987). The skyrocketing sales of the iPhone led to some tough decisions. Apple no longer needed to control the whole ecosystem with their server product and instead began transitioning as many teams as possible to work on higher profit margin areas, reducing focus on areas that took attention away from valuable software developers who were trying to solve problems many other vendors had already solved better. In 2009 the Xserve RAID was discontinued and the Xserve went away the following year. By then, the Xserve RAID was lagging and for the use cases it served, there were other vendors whose sole focus was storage - and who Apple actively helped point customers towards. Namely the Promise array for Xsan. A few things that were happening around the same time. Apple could have bought Sun for less than 10% of their CASH reserves in 2010 but instead allowed Oracle to buy the tech giant. Instead, Apple released the iPad. Solid move. They also released the Mac Mini server, which while it lacked rack and stack options like an ipmi interface to remotely reboot the server and dual power supplies, was actually more powerful. The next few years saw services slowly pealed off the server. Today, the Mac OS X Server product has been migrated to just an app on the App Store. Today, macOS Server is meant to run Profile Manager and be run as a metadata controller for Xsan, Apple’s clustered file system. Products that used to compete with the platform are now embraced by most in the community. For the most part, this is because Apple let Microsoft or Linux-based systems own the market for providing features that are often unique to each enterprise and not about delighting end users. Today building server products that try to do everything for everyone seems like a distant memory for many at Apple. But there is still a keen eye towards making the lives of the humans that use Apple devices better, as has been the case since Steve Jobs mainstreamed the GUI and Apple made the great user experience advocate Larry Tesler their Chief Scientist. How services make a better experience for end users can be seen by the Caching service built into macOS (moved there from macOS Server) and how some products, such as Apple Remote Desktop, are still very much alive and kicking. But the focus on profile management and the desire to open up everything Profile Manager can do to third party developers who serve often niche markets or look more to scalability is certainly front and center. I think this story of the Apple Server offering is really much more about Apple branching into awesome areas that they needed to be at various points in time. Then having a constant focus on iterating to a better, newer offering. Growing with the market. Helping the market get to where they needed them to be. Serving the market and then when the needs of the market can be better served elsewhere, pulling back so other vendors could serve the market. Not looking to grow a billion dollar business unit in servers - but instead looking to provide them just until they didn’t need to. In many ways Apple paved the way for billion dollar businesses to host services. And the SaaS ecosystem is as vibrant for the Apple platform as ever. My perspective on this has changed a lot over the years. As someone who wrote a lot of books about the topic I might have been harsh at times. But that’s one great reason not to be judgmental. You don’t always know the full picture and it’s super-easy to miss big strategies like that when you’re in the middle of it. So thank you to Apple for putting user experience into servers as with everything you do. And thank you listeners for tuning into this episode of the History of Computing Podcast. We’re certainly lucky to have you and hope you join us next time!