So I’ve had this thought for some time now, and have shared it with a few people here and there. Before I start getting deep into speculation, I want to lay down some facts:
- Oct 2010 – Apple outsources Enterprise sales and support to Unisys.
- Nov 2010 – Apple announces the discontinuation of the XServe. At the same time, they release a laughable white paper, referring to the Mac Pro and Mac Mini Server as reasonable replacements. The Mac enterprise community is shocked, not so much at the discontinuation of the XServe, but at the options left behind.
- February 2011 – Apple releases Light Peak (rebranded as ThunderBolt) enabled Macbook Pros. ThunderBolt represents a huge increase in external interface bandwidth for consumer-level equipment, allowing them over three times the bandwidth of the fastest interface- eSATA.
- April 2011 – Promise announces SANLink ThunderBolt to Fiber Channel interface, with assistance from Apple in the design.
Macs in the data center.
Many Mac Administrators have begun to believe that Apple is just giving up in the enterprise market, relegating the server space to Windows and Linux machines running on VMWare virtualization clusters of IBM/Dell/HP blade servers.
My thought is the opposite. Apple can make a compelling case to replace the blade servers from other manufacturers with Apple kit.
The following is entirely speculation, utilizing the facts above, as well as past rumors that have proven true.
In 2009 or early 2010, Apple makes a decision to replace the XServe, an aging platform that has been usurped as the cutting edge some time ago, with the Mac Mini. This is in line with Apple’s core skills of miniaturization, as they have talked about in the iOS-based and MacBook Air devices. To execute this plan, they need to re-tool the XServe factories, as well as shift the XServe engineers over to other departments, which naturally takes time. They do so immediately, knowing full well that this means that they can no longer sell XServes in 2011.
At the same time, they are working with Intel to bring a very high-speed interface to small devices, such as the MacBook Pro and Mac Mini. With speeds starting at 10 Gbps and expected to top out at 100 Gbps, ThunderBolt is a technology with legs. They plan on implementing a Mac Mini Server with at least one ThunderBolt port, as well as lights out management, with an intent of having it ready be announced with the discontinuation of the XServe, and ready to ship in Q1 2011.
Unfortunately, there appeared to be last-minute issues with the ThunderBolt interface which caused the updated Mac Mini Server to be delayed. Due to the production changes made months earlier, Apple was still forced to drop the XServe. Without a valid replacement, Apple releases the white paper to the public, and waits to release the Mac Mini Server.
The Next Mac Mini Server
I think that a ThunderBolt-enabled Mac Mini Server is not really even speculation. What they do with it is. Here’s a list of what I think makes the most sense:
- Replace the HDMI port with a second ThunderBolt Port. Replace the Firewire port with a third.
- Provide Lights out Management.
- Provide VMWare ESX support on the Mac Mini Server.
- Provide a blade chassis that supports the Mac Mini Server as the blade.
- Create, or partner with 3rd parties to create, a line of enterprise-centric ThunderBolt break out boxes, such as the SANLink. I am thinking things like multiple gigabit ethernet ports, eSATA breakouts, legacy Firewire.
These items will create a low-cost machine that can be used in the data center in exactly the same way that Google and Facebook do so – low cost, with redundancy in the machine, not in the component. But most importantly low power consumption. The Mac Mini Server only consumes 85W of power. This is compared to over 500W for HP’s most efficient blade server the BL260c G5, and in line with Facebook’s recent open computing specification, which is a custom piece of hardware.
Now, imagine a world where a data center could provision equipment directly from the stock of the local Apple store. No need to store costly extra power supplies, hard drives, logic boards, and fans. Instead, you have a spare Mac Mini Server or two, and all it takes is a trip to the Apple Store if they fail. The cost of the machine results in the ability to more easily create redundant machines. Compare this to the cost of ActiveStorage’s ActiveSAN, which clocks in at a whopping $23,000 for the baseline pair of machines, and you can have twice as many Mac Mini Servers with internal SSD hard drives, XSan, and SanLink (who’s price has not yet been released.) And that’s not taking into account Apple’s ability to lower the cost of the SSD hard drive by purchasing in mass quantities.
Of course, even if I am correct in the above suppositions, Apple has a history of dealing poorly with the enterprise. The lack of product roadmaps result in many managers reading tea leaves, as I have done in this article. In fact, Steve Jobs himself has come out and said that Apple basically ignores the enterprise market. He also indicated that there would not be an Apple phone, video iPod, and many other products that have come to market. However, there is one indication that Apple is starting to become serious about the enterprise.
As I mentioned at the start of this article, Apple has outsourced their Enterprise sales and support to Unisys. However, Unisys’ business is all about the enterprise. The work within government and data centers. They’ve been in the enterprise space since nearly the beginning of computer. They have a position that will allow Apple to build their credibility in the enterprise in a way that Apple cannot do alone. They’re building the halo effect now with the iPad and iPhone, then when the Mac Mini Server is released, look out.
Anyhow, that’s my thought about what the last several months of action from Apple and their partners mean for Mac-based enterprise. Please pick it apart in comments below.