Beta Testing with the Apple Server Group
These notes about Chuck Goolsbee's beta testing experiences are taken verbatim from my correspondances with him.
I was a participant in the ANS testing from late 95 through release in early 96. Frank Nicolo was the (hardware) product manager. Frank came to Apple from DEC. His project was to build the world's best server enclosure. I think he lived up to that task quite well. Unfortunately for Frank he worked for Morons. Frank left Apple for who knows where?
My only involvement with the AWS95 prior to its release (when my employer bought one) was being one of the first customers to see one at Mactivity94 in San Jose. Marv Su was the product manager and I was one of the first 50 people to see the AWS95. We were supposed to see an actual running model, but that was *stolen* from the exhibit hall about 15 minutes before "showtime!" The sight of people rolling computers around on carts at trade shows is not unusual so nobody really took notice to the guy who walked off with the very first fully functional AWS95. Marv, after a moment of abject terror, managed to pull off an excellent demo. I recall that he "made" an AWS 95 on the spot by taking a Quadra 950 brought up from Apple's booth and installing the PDS card and A/UX. I walked away with an appreciation for Marv and an "Apple Workgroup Servers" painters cap. Marv left Apple in 1995/6 for eXite or Lycos or somesuch internet search-engine startup and is now probably a bazillionaire.
I tested the ANS 500 & 700. I had a 500 from November 95 until December 95, and a 700 from December 95 through early summer 1996 when I was asked to return it to Apple. It was released in early 96 and was featured prominently at MacWorld Expo San Francisco that year. Unfortunately Apple did not really know what to do with a 'real server' once they had one. Only one of the servers on the show floor had a monitor on it running AIX and quite a few were running the hacked version of MacOS that was made to run on it (sort of). This was of course a completely mixed message to send to the attendees. The box was never designed to run MacOS - ever. At best it would be on par with a 9500. It was tuned to run AIX, and it did that very well. As a Macintosh it would be a bundle of missed opportunities like the IIfx: tons of cool hardware that would never be supported by the OS and as such be not worth the hefty price tag. I remember being quoted in MacWEEK. Big time geek-points for that back then. I recall a chance meeting Frank Nicolo at the gym in the ANA hotel at MacWorld 96. He arranged a dinner meeting between myself and his new boss. This guy represented all that was wrong with Apple during their darkest days. His name was Baker. What a moron. He was this total (no offense!) east-coast, Ivy League, MBA type. The words "pompous" and "indecision" come to mind for some reason. After that meeting I had a strong suspicion that the whole project was doomed right here at the start. With somebody like Baker at the wheel Apple's server group was like the Exxon Valdez: headed for the rocks...
The code name for the ANS 700 was "Shiner" (named after a beer brewed near Austin Texas, home of the development team).
At the time (1993?) I first saw the early pre-Shiner (ANS) demos by Apple of Netware and AIX running on a weird PPC 601 in a Quadra 950 case prototype, I was working for The Bon Marche, a division of Federated Department Stores. This was at an ANMA meeting at Apple's now defunct Market Center in Kirkland WA. I'm pretty sure Frank Nicolo & Robert Patrick (aka "RP") were there from Cupertino. I recall the Apple host as being Kurt Gotfredsen who now works for AT&T wireless.
I was doing some serious OPI (Open PrePress Interface) serving to a huge room full of Macintosh production workstations using then state-of-the-art Sun Microsystems Sparc servers and huge(!) MegaDrive MR/5 drive arrays that held a whopping 8 gigabytes in a raid-5 config. That was a mind-boggling amount of storage at the time since the average desktop drive was under 100 megs.
The two guys from Cupertino were all excited to show us (about 15 ANMA network geeks) Netware ported to PowerPC. We said "no thanks" to Netware and said "show us your UNIX." They were floored, since at the time Netware was *the* NOS of choice but we knew its days were numbered. These were the days of John Sculley & Michael Spindler and promises of "Pink" and "PowerOpen." We never saw Netware beyond a startup screen because we didn't want to.
Of course we didn't see UNIX then either, but the Apple guys went back to Cupertino after sucking our collective brain dry for hours about what we wanted in a server.
They came back in late94/early95 with the "Shiner" which later became the ANS. It was a very impressive machine. I was signed on to the beta program. What was weird is that right after I started the beta I switched jobs and left The Bon Marche to go work for TPD Publishing... and I took the beta server with me! TPD was not really a great testing arena for the box. The Bon was perfect in that it was high-volume advertising with about 140 users, whereas TPD was a small startup with about 15 people on staff and only one project... ironically "Microsoft Magazine" (!)
I used the ANS for some internal testing, and helped de-bug a RAID card and had about 10 people using it as an applshare server. I was not able to put it through the paces originally expected of me but I did help shape the box's software and user interface a bit. I think mostly I helped evangelize my peers in the publishing business. Microsoft Magazine's Fall/Winter 1995 issue was produced using the beta Shiner as its server for some of it's images. All design & production for Microsoft magazine was done on Macs.
Things were cancelled in piecemeal. I think Frank Nicolo had already left and RP was running the program when the Deep Dish project started. Baker was still at the helm of the server group, unfortunately. They were actually making money, due to higher margins, but was obviously not cranking out high volumes. The ANS was never marketed very well at all though and never did sell very well compared to the MacOS based servers. I think this had a lot to do with Apple's deep internal schizophrenia about UNIX. Apple, like any large company, suffers from extensive "NIH" problems. "Not Invented Here" prevents people from pursuing projects that otherwise would succeed if given the chance. UNIX was invented before Apple existed, and many within Apple have always felt and acted like it was some sort of evil monster that was too scary for their customers to handle. The reality was (and is) that many of their customers had handled it fine for a long time and were already using it because it was and is a superior server platform for many tasks. Why cede that market to Sun, SGI, HP, etc.?
Ironically Apple was in search of a modern operating system to replace, or merge with the MacOS and ended up buying a UNIX company in the form of NeXT. Additionally it was the merger of the two companies that ended the ANS program. With a future being placed on a different UNIX and a hardware line that was being pared down from the myriad models of 1997 to the four basic lines we have in 1999. Apple's entire server group seemed to vanish shortly after Steve Jobs returned to Apple. In many ways I understand... Apple had to cut things back to basics and return to profitability. As a customer and a stockholder I would prefer to see them selling millions of iMacs than thousands of servers. On the other hand as a network administrator who has built and retired many servers, Apple's withdraw from the market left most of us holding onto instantly obsolete machines with little or no future, or running to the few alternatives out there. WindowsNT captured a huge share of the Apple server market with an absolutely crappy product in it's "Services For Macintosh"... the "free" NT afp server. Egads what a lousy product! But it was one of the few alternatives during this time. Linux was not yet mature, and the other UNIX afp servers were so expensive (many with big buck per-seat licensing.) Apple continued with AppleShare, finally adding IP as the transport protocol with AppleShare 5, but so much in the MacOS kept it from being a big winner. Stability primarily with the running disaster that was 7.5, and other issues such as the aging file system and so-so I/O. (lousy I/O compared to UNIX or NT really.)