I want to believe
Written by marc on June 15th, 2011
Even a casual observer of Apple in the enterprise will agree it has been a depressing year with the spontaneous assassination of data centre hardware, continued head-in-the-sand attitude to virtualisation and a further year of veritable neglect to its server software. The industry has waited for a year to find out the just-announced Lion Server contains little new substance and that there is still no hardware to run it on.
Wait a minute. How did we get here?
I was living in Canada during that amazing WWDC 2002 moment when Steve finally got serious about back end hardware and software and introduced Xserve and Mac OS X Server. I was fortunate enough soon after at a high level conference at the snowy Kananaskis Delta Lodge, Calgary to meet both the hardware and software product managers for Xserve. Both were ex-Sun and spruiking big product lineup plans for the future. The 1RU form factor combined with dual PowerPC G4 gigahertz processors was at the time the best price per MIP in the industry and was, for the first time, data centre friendly. The later G5 version even resulted in remarkable installations like the Virginia Tech System X supercomputer which briefly ranked #7 in the TOP500 list.
I admit that I drank the kool-ade and, for once, even the press loved Apple's back end offerings. Rave reviews of the debut product line continued as Xserve evolved into the later 2RU G5 form factor and Xserve RAID. Mac OS X Server 10.2 'Jaguar' set a new benchmark for ease of server setup and configuration (that is at least until something went wrong with it) and the unlimited user pricing was simply disruptive. For a while there the horizon was looking good - for good reason.
The industry love began to fade however as Xserve began to be left behind as the trends towards virtualisation and power efficiency saw the industry shift to blade servers and VMWare and its ilk. By the time the Intel versions of Xserve showed up, Apple was so far behind the band in the data centre, it was like a replay of the disasterous Apple Workgroup Servers.
Since then these devices continued to lose ground and increase in irrelevance. A slow decline in product commitment began by strangling Xserve RAID back in 2008. Underinvestment and underdevelopment continued until finally in February this year Apple committed seppuku on the entire Xserve line.
However, distressingly for any Datacentre Operations manager, the company failed to do the honorable thing and exit presence in the data centre altogether. Mac OS X Server remains "undead" with no server class hardware to run it on, and Apple continues to this day to push its unfortunately named "Golden Triangle" combination of Open Directory and Active Directory to best manage Macintosh in the enterprise. This situation would be comedic if it wasn't so tragic.
Open Directory is laudably open in many respects, but in the face of the ubiquitous defacto standard Microsoft Active Directory, Apple's implementation is effectively proprietary, un-virtualisable, and requires specialist skills that are impossible to find. Career prospects aren't great for Macintosh enterprise people given the uncertainty of this whim-of-Apple market.
The Ops manager is left in a situation where, just to support freedom of desktop operating system choice, they must run Apple's unfamiliar and unique server software on workstation class machines and cannot virtualise either hosting (no VMware) or access (no built in terminal services).
So, gentle reader, you may wonder what the point of this post is?
At WWDC, after five months in limbo Apple finally confirmed that Server has, for now, a viable future by announcing Mac OS X 10.7 Lion Server as a US$50 downloadable add-on to the Workstation operating system. You have to wonder who this is targeting. Pricing was already aggressively good for Mac OS X Server with unlimited user licenses - a stark contrast to Microsoft's server and client access licensing. A single webpage of product info and a cheap sticker price won't increase Lion Server adoption in the enterprise data centre.
Oddly we see random and disconnected green shoots in the various Lion announcements that offer some hope that somebody is still employed in Apple Enterprise Product Management, including:
- NFSv4 support
A tantalisingly small step away from the Parallel Network File System implemented in v4.1
- SMB with DFS
Support for Windows Distributed File System URLs, drill-down, failover, and reconnection to Windows file servers.
- Profile Manager
Built in implementation of iOS Mobile Device Management APIs
- XSAN bundled
XSAN, Apple's implementation of the Quantum StorNext cluster filesystem, is now included in every copy of Lion workstation and the host software is bundled into Server.
Rumours persist that we may see a new form factor computer Real Soon Now that will satisfy both deskside and rackmount configurations. This might just serve the SME market, but will not remotely satisfy green standards in datacentres which specifically relate to not only core density per rack unit, but virtualisation and utilisation of those cores.
The combination of all these strange tidbits is simply inadequate as a strategy, let alone a roadmap. My appeal to Apple is "get in the game or get out of it altogether because what you're doing right now is akin to radiation poisoning of Macintosh in the enterprise".
I say: either discontinue Server altogether and redirect resources to... oh, say make SMB and native AD client-only implementations work properly; or allow Server to be (legally) virtualised in its EULA to allow it to be hosted within VMware on true datacentre class hardware. A third option is to get back in the game with blade hardware, but nobody should hold their breath since this would be both regressive and anti-iCloud.
Macquarie is not alone. In 2010 I canvassed CAUDIT, the Council of Australian Directors of Information Technology and found current mainstream Australian university Apple customers would be more predisposed to purchase Mac OS X computers if there was a server offering that could be virtualised with VMware (or Xen or similar) and support terminal server style multiuser access via Citrix (or VMWare View or similar). Apple has had this data for a year and is yet to officially comment.
Ironically, when I started in higher ed I thought that it would be a seachange to enjoy core business attention by Apple Inc. Frustratingly, Apple it seems has transcended that traditional home ground and moved on to direct consumer relationships with its iOS strategy, meaning that even mainstream universities experience indifference. Apple risks losing all lab computing across the board in most universities, and relegating students and staff to a second class personal experience by making management of Macintosh a tenuous and minority pursuit. Ultimately that threatens freedom of choice at the desktop - something I'm passionately committed to.
Don't get me wrong, nobody could be unhappy with the leadership and consumer tech at competitive price points that Apple is delivering these days. I want to believe that it is not willing to sacrifice heartland Macintosh adoption to do so.
Hi
From and had done bit on the server Cant understand what Apple plans r for server. A neat energy efficient server along with OS cando the trick.
I spotted this article from Macsurfer.com and fully agree with the sentiments expressed here . It saddens me to hear that Apple is demoting the Mac and now seems fully focused on iOS . I guess Apple telegraphed this when they changed from Apple Computer to Apple Inc .
To depress you more check out this quote from cringely.com :
“What this requires from Apple is a bold move that Microsoft would never make: Jobs is going to sacrifice the Macintosh in order to kill Windows. He isn’t beating Windows, he’s making Windows inconsequential.”
http://www.cringely.com/2011/06/iclouds-real-purpose-is-to-kill-windows/
Cheers : Kevin
I agree – Apple should have never killed its XServer hardware … it was building a good business, even it was tiny and not as profitable as they wanted. OS X lends itself to a server class OS. I think if Apple was smart they would have scaled their OS X Server hardware and software to their cloud centers. I know Sun has a longer track record running data center OSs and there is this ‘huge’ cloud thing come up.
I would think even if the cloud is as successful as Apple dreams, there is still plenty of space for real XServe hardware – to manage systems in business … graphics shops … audio shops… schools and everywhere else. There are a lot of companies that would not put anything in the “cloud” not matter how cheap/reliable/efficient it is, because they have confidential data and they need to control this … secure and on site. Apple is really dropping the ball here.
The ironic part is that Apple is dropping the hardware (and probably the server all together?) because the iOS has taken off so much. In other worse the XServe and OS X Server fail because Apple succeeds … WTF!
The long and the short of it:
you should have bought more servers.
Money talks and bullshit walks. I’d bet real money that more people bought the MacMini with server bundled in the first year it was out than bought Xserves for the previous 5 years before it. The profit margin on a software package for an additional $300 was probably greater than they ever made on the Xserves, on a per unit basis.
My guess is that the sales of Xserves was anemic at best, and failing at worst.
But no. Datacenter folk want to buy less hardware, use more virtual servers. Buy fewer storage systems and just add dirt-cheap disks to an existing pool.
If you all would put your money on the table, Apple’d make something you’d want. Not “promise to look into it for Q3 of next year” or “consider buying when we get a 2 year roadmap from you, Apple.” You could have bought stuff. A lot of stuff. Then Apple would have thrown more money and engineering behind it.
Right now, Apple would rather make $50/yr from 5 million people, than $50,000/yr from 200 datacenters and enterprises. The 5 million people are more than willing to pay and have.
@N Agree – I don’t weep for Xserve and have no interest in its replacement but ignoring virtualisation for Mac OS X Server is a much more bitter pill to swallow. Mac OS X Workstation will never be virtualised and I completely understand that position.
@Moe It’s hard to get upset about the death of Xserve – a quick look at Cisco UCS or the HP equivalent and there is just no comparison in hardware bang for buck or energy efficiency.
Cloud is a whole other argument, and actually I’d be over the moon to see Apple say ‘resistance is futile, we’re killing Sever entirely’. Trouble is they aren’t doing that – they’re half pregnant by attempting to play in a game they are not committed to.
Wouldn’t you just die to know what hardware and software underpins iCloud?
Thanks for the comment @Kevin. I figure iOS == Mac OS X, just a custom distribution in reality. I think of it as the new batch.
There is hope!
http://www.appleinsider.com/articles/11/04/30/vmware_vsphere_5_to_add_cloud_virtualization_support_for_mac_os_x_server.html
We’re in the beta programme for Vsphere 5. So far, this doesn’t appear to be accurate though I very much hope it’s true.
Dang. It seems the most sensible option.
That or vendors offering enterprise grade systems certified for OS X Server by Apple.
sad and:
“Wouldn’t you just die to know what hardware and software underpins iCloud?”
yep!
[...] authentication and directories, to virtualisation, it appears Apple just can’t get it right. Writes Bailey, in a post entitled “I want to believe”: “My appeal to Apple is “get in the game or get out of it altogether because what [...]
So basically you’re telling Apple “Copy everyone else, or get out”.
Not going to happen because:-
A) Apple lead, not follow. With enterprise IT clocking a 90% failure to deliver any discernible ROI, they definitely don’t follow.
B) Legitimising virtualisation of OS X will effectively transition it to a licensed model which would not only destroy the Mac’s business model but the tight HW/SW integration that makes the Mac a Mac, even a server.
C) The server OS is dead and will be rapidly replaced by OS agnostic cloud services. Apple have, cleverly, sumised all that remains is the SMB workgroup/basic directory & print services running on today’s over-adequate Core i7 desktops.
Re-iterating point A, perhaps it’s not Apple who need to change.