Xserve End Of Life: Some opinions and ideas about Apple's server strategy

Apple has been getting away from enterprise for a long time. XServe RAID was slow and outdated the day it debuted - frankly i'm surprised anyone bought the thing in the first place.

The XServe was a decent machine, but suffered from lack of interest (from Apple AND potential buyers).

Mac OS X itself is more interesting - It's half extensions of Apple Software (iCal server, Address Book server, WebObjects, etc) and half implementations of OSS. Using OSS was an easy and inexpensive way to get a server product together, and it wouldn't shock me at all to see Apple drop those components and allow people to install them on their own. MySQL, PHP, Apache all of these have free third party built Mac OS X implementations that are better than Apple's. That would allow Apple to concentrate on (and modularize) it's proprietary apps - a better solution all around.

It's pretty clear that Apple doesn't want to be in the datacenter in any major way - the only role they want their is to directly extend their end user experience. Let everyone else manage the other stuff.

http://www.xserve-petition.com/

The Xserve is a bit of a painful form factor. 1U server made sense in a world where ISP's and Co-Lo providers charged by rack space. Those days are largely gone. They now charge by power consumption which ties directly to cooling and delivering high power capacities to individual cabinets. Putting 15,000 va into a rack is incredibly hard to cable, power and cool.

Most places are happy just to give you another cabinet.

The thin long servers just forced everyone to go buy 4 ft deep racks to accommodate all the long servers that jammed up the backs of your cabinets with cables, blocked air flow, and prevented you from shutting the cabinet doors.


I'd be very happy if they modified the Mac Pro a little bit.

The removable XServe drive trays should stay. They are wonderful.

Mounting an XServe requires a traditional cabinet or a 4 poster.
They don't go on a shelf or desk nicely, and they don't fit in a traditional relay rack.

Change the Mac Pro so that you could take the handles off, and attach rack mount ears to it, and mount it sideways in a rack.

A 19" form factor Mac would make an awesome Home Theatre PC, Small Company Server, or desktop machine.

See ATech Fabrications cool cases if you want an idea of what a Mac Pro server could look like.

http://www.atechfabrication.com/products/HeatSync_7000.htm

(No, I don't work for them, I don't even live in the same country. I'm just lusting after a Mac Pro logic board mounted in their HeatSync 7000 Case)

I don't think the end-of-life date of January 31, 2011, is an accident. It's right after MacWorld.

My guess: Apple decides to license some hardware manufacturer (like HP) to make OS X Server-compliant server hardware, and then Unisys would provide enterprise support the way they're already signed on to provide enterprise support for the iPad. In this scenario, Apple designs the motherboard & other hardware specs but leaves production to someone else. Apple verifies compliance of the finished hardware so everything works 100% with OS X Server. It's a win-win-win: the enterprise can breathe a sigh of relief because they're used to dealing with HP and Unisys anyway; HP and Unisys get more corporate customers; and Apple pushes OS X deeper into the enterprise with minimal effort and commitment of resources.

Everyone forgets that Apple is a *hardware* company and that everything else -- including OS X Server -- exists to sell more Apple hardware. The problem with the XServes is that it was costing Apple too much to support them. By offloading the *server* hardware sales and software support to other companies that are more known and trusted in enterprise circles, it makes it easier for Apple to sell tons more *desktop* hardware to the enterprise. What makes Apple more money: selling (and supporting) 2 Xserves, or selling 200 iMacs to outfit a whole company or department?

At first I was a bit shocked, but then my more reasonable daemons prevailed. I don't think Apple is moving away from the enterprise. I think they are moving towards it.

I would hazard to guess that the Xserve and the MacPro have the lowest numbers of all of Apple's form factors. Neither has really changed since their introduction in 2002 and 2003. That's a long time for Apple. The original Xserve was the wedge to enter the enterprise market. In the current market where virtualisation is king and SAN's prevail it's no longer relevant. Mac OS X Server has three major customers: small businesses, enterprises with existing infrastructure, or FCS/XSan customers. The Mac mini server nails the first two,

For the remaining customers the MacPro is not good enough. I bet that an overdue refresh of the form factor is on the way that allows for a vertical or rack-mounted horizontal orientation. It may have options for internals to match the Xserve's credentials. Or following in the footsteps of the MacBook Air, there will be a blade server running on flash memory that will truly be high performance. They've put SSD into every other form factor.

As an Apple consultant, I've noticed Apple lifting their game over the past few years in terms of support and promotion. You don't provide podcasts on server technologies or evolve your branded consultant's network if you are planning on selling just desktops. I think is part of a long term plan to hone their offering as the go "Back to the Mac". Unfortunately for system admins it is part of Apple's DNA to hide the roadmap and have items magically appear on a stage.

Of course I could just be drinking my own Kool-Aid….

What about they keep MacOSX Server and qualify it to run on certain other Intel server hardware? They still charge quite a bit for OSXS so they would still make a good margin.

While I'm jumping into speculation, I might as well throw this out as well. Apple has already doubled the size of their server farm in N. Carolina before announcing it's role. Suppose that Apple releases a virtual client on iOS, and they host cloud based osx virtual boxes you can access from an iPad. This makes the iPad/iPhone a very portable thin client. I could see a few things coming together to make this happen.

The only way this move makes sense to me is if 10.7 will become available for virtualization. As a MaxOS Server manager, I think it's very risky for Apple to announce the death of the xserve without a viable high density enterprise option. The only conclusion (guess) I can come up with is that there will be licensing options for running OSX server in a ESX scalable virtual platform, or on other intel hw.

I can see Apple eventually allowing OS X Server in a virtual environment on any x86 server (VmWare ESX..etc) There really is no need for Apple's industrial design when it comes to servers.

Well as far as I am concerned, apple is abandoning all but the smallest business users and home users.
We support many schools, most of which have and need xServes. I dread the idea of running our school systems on MacPros but what other choice will I have?

I too agree that Mac OS X Server is disappointingly buggy, but if Apple stopped spending all their time playing with their shiny iPads and iPhones then they could probably round up most of the bugs and get them into touch.

So, Mac's managed by a Windows Server... I think not. I guess I might as well put my Mac in the bin and go and buy a PC.