Failure Analysis: How Apple Screwed the Xserve Pooch

| Hidden Dimensions

“Failures are finger posts on the road to achievement." -- C.S. Lewis

Many Apple customers throughout the years profitably utilized the rack mounted Xserve product in their businesses. It was a quality product that cost less than the competition's by a significant amount. Yet, it was doomed to failure. Here's how it happend.


Buried in an article about Android, Apple and market disruption, on page 2, Daniel Eran Dilger describes, briefly, the failure of Apple's 1U rack server, the Xserve. The article is: "Apple's market disruption savvy is bad news for Android."

The main thrust of the article is about the art of market disruption and how observers should view Apple's iPhone strategy. Good in itself, by the way. What caught my eye, however, was that innocent looking section on page 2 about the Apple Xserve.

A Lesson in Disruption

As part of his treatise on Android market disruption, Mr. Dilger uses the example of a failed market disruptor, Apple's own Xserve.

When the company tried to break into the enterprise server and storage market with the 2005 Xserve and Xserve RAID, it attempted to leverage its Mac OS X Server platform to deliver equipment that could rival the higher end offerings of rack mounted server room equipment.

Apple hoped to undercut established PC server vendors, including Dell and HP, by offering a server package that delivered comparable hardware with vastly cheaper software than Windows Server, resulting in a much lower price overall."

A comparison chart is shown that was used in Apple marketing to demonstrated the real, tangible cost savings when using the Xserve. Many, many customers bought into this, especially those who were able to exploit and manage the Xserve themselves for effective use.

Very impressive.  But it didn't work in many cases.

In one sentence, Mr. Dilger correctly summed up the ultimate failure of the Xserve:

Apple's Xserve never established much of a market for itself, in part because the company lacked the hands on support services that enterprise-savvy vendors like Dell provided to their satisfied customers.... "


When I was at Apple, in the Federal Sales group from 2003-2005, I was selling the Xserve and its storage companion, the Xserve/RAID. I can confirm what Mr. Dilger wrote above. In my own words, I would say that enterprise/federal customers didn't want a relationship with a product. They wanted a relationship with a company, and they were willing to pay for that. On call.

Apple Xserve (2002-2010)

Apple, on the other hand, was of the mind that products should, by their elegance, sell themselves. That was the Steve Jobs vision. No formal, tedious relationship with the customer should be necessary. Yet enterprise customers could be vastly annoying; they were always whining about what they wanted and needed. They demanded insight into future changes.  They'd ask to be involved. Apple senior management never felt comfortable going there, and so many customers with ample cash fell into the arms of HP and Dell for their servers and support. In that sense, Apple failed.

By and by, that disconnect resulted in limited opportunities for the Xserve. The big opportunities for Apple, the low hanging fruit, lay instead in the consumer electronics arena.

This goes a long way towards explaining how the Xserve could be, at once, both loved by many customers and yet be relegated to the museum of failed Apple products. Customers of mine were always amazed: how could Apple build such a great product, then fail to support it in the way that they needed? (To be fair, Apple field engineers, far too few and scattered, did a heroic job of sales support.)

A good example from those days was a major military organization that needed assurances from our team that Apple would manufacture the PowerPC/G5 Xserve for years to come -- right at the time when Apple secretly knew it would be shifting to Intel. Selling to the military (or trying to) could be a bitch that way. Senior Apple management elected to defer on a potentially huge sale for the sake of getting on with technology, never allowing itself to be held back.

As an aside, that inability to put technical and leadership support behind computational clusters led to Apple's departure from supercomputing as well, but that's another story for another day.

We tend to think of Apple as a company that can't do very much wrong. However, in this case, Apple was in the process of struggling to find the right market profile while also understanding that its way of doing business didn't fit in, in critical ways, with enterprise America. One way to view it was that Apple put its toes in the water, blinked and eventually withdrew, leaving some scars behind.

That's not to say Apple didn't have many, many satisfied enterprise customers, large and small. And it's still true today in the enterprise and government. Plus, Apple's focus on small business has been legendary. Apple made millions of dollars in Xserve from 2002 to 2010. But customers just kept wondering: why couldn't the best hardware company on the planet go all in with them? And the Apple sales teams, straining mightily to meet their sales quotas for an admittedly great product, wondered the same.

Eventually, the company found that it was self-limited in what it could do there and discontinued the Xserve in late 2010. The Xserve era was a great ride, but, in the end, Apple had to screw the Xserve pooch, dust itself off, and become the iPhone/iPad company. Fortunately, Apple found its true niche with consumers. And so, it put an experimental past out to pasture and hasn't looked back.

Who knows what Apple might have achieved if it had approached enterprise sales in a more conventional fashion? But other companies, with the right name recognition, did it better and Apple didn't see a big future in competing with them. And so, it's hard to argue that, given Apple's unique penchant for product design and approach to the modern consumer, the current path is a bad idea.

Postscript on Xserve/RAID

While Mr. Dilger is almost always on the mark, one comment he made about the companion storage product, the Xserve/RAID, didn't ring true. His analysis:

XServe RAID was canceled even earlier because the NAS market wasn't at all impressed by lower prices; they preferred the assurance of acquiring drives rated for true enterprise duty, rather than the consumer-class disks Apple was trying to sell them as cheaper and "good enough."

I sold the Xserve/RAID myself at Apple, and I was even accompanied on a sales call at a national laboratory by Mr. Alex Grossman, the Apple executive behind that RAID product. He would explain to customers that the resiliency of a RAID product is a function of the entire system, and that includes the RAID controller.

Xserve/RAID (2003-2008)

As background for this story, we chatted again, and he reiterated what I recall us telling customers: "The drives in Apple's Xserve/RAID were perhaps slower than the ultimate Enterprise-class drives available at the time, but as an aggregate, that product performed very well. Not only did no customer ever lose any data by virtue of the design, but no customer ever complained or deferred purchase because of the drives we used."

Instead, despite stellar sales, Mr. Grossman (now at Quantum) told me that the Xserve/RAID also fell victim to Apple's "Opportunity vs. Engineering Cost" analysis. That is, given limited engineering manpower and resources, a whole lot more could be achieved by applying those resources to iPhone. Xserve/RAID, while a beloved and spectacular product, would never begin to achieve the potential sales, in dollars, of the iPhone. It was discontinued in 2008.


There was a time when Apple was a boutique UNIX company selling Macs to all kinds of technical professionals. It was natural to back that up with RAID/SAN storage products, Xserve, and other enterprise-class products. Customers loved everyone of those products, but remained baffled by Apple's customer support approach in general. In the end, those products weren't the path to the kind of eventual success Apple wanted and was uniquely suited to deliver.

As some have noted, it is, however, a shame that so many technical and enterprise customers who have adored all kinds of Apple products for years got left behind in that product space. Perhaps the frustration was due to wish fulfillment -- that Apple could be both great at both hardware and traditional business support.

We tend to think that a very large, wealthy company can continue to do all things well. Certainly, Apple could have invested vigorous resources into securing an important market for itself in servers, storage and clusters. That temptation to do everything well can often lead to not being spectacular in the areas that really count. Or working in areas where executives feel uncomfortable, out of their depth.

In the end, Apple chose to think different. It matched its finite engineering resources to the market best suited to its own legacy, temperament and vision.

It seems to have worked out fairly well.


Pooch image via Shutterstock.

Popular TMO Stories



Very interesting article.
A high powered niche product that the users love but takes a lot of engineering time for decent but not earth shattering money. A product that is customized to fit each customers particular needs? A product that then gets put under a table or in a closet and is never seen by anyone except the user, unlike the iPhone which is a walking ad for itself every time a user pulls it out? Could it be in a couple of years you will be able to dust off this piece, replace Xserv with MacPro and publish it again? I see a lot of parallels.


Were there ever any companies that would fill the customer support role by reselling and supporting Xserves? Is that not good enough since they couldn’t speak to product direction?

I would ask whether a company could start selling Xserve-like hardware to put OS X Server on and then support it. This would allow the company to give commitments to hardware if not software. But then this would be an Xserve clone and we see what Apple does to clones.

So, last try. Could/does some company do the same model with Linux as the software? (That model being comparable quality hardware at comparable cost but much cheaper software licenses.) HP et. al. do offer Linux servers, are those at this cheaper software license rate? Is the catch that there’s no good Linux competitor to MS Exchange?

I’m not against paying for good products, but I’m very much against paying a per-seat fee since it imposes arbitrary limits that are not tied to production costs and only serve to provide profits. Though I guess more users means more support calls, but that’s not always true and I’m not sure the model in the chart above accurately matches license prices to support costs. (E.g. does going from 10 to 20 users actually double support costs?)


I’m not against paying for good products, but I’m very much against paying a per-seat fee since it imposes arbitrary limits that are not tied to production costs and only serve to provide profits.

For example, my company has decided to install an exchange server. Not the solution I wanted to go with, due to cost, but the powers that be made a call.
Hardware: $5000
Total cost of the system with Windows Server 2012, Exchange, and per seat licenses: $30,000
The killer is the licenses. It’s truly obscene what MS charges for them.


Apple thinks it’s clever to shave off customers.  1% here 1% there, no big deal.  Maybe all very justifiable.  However I note that apple lost 1/3 of its value and the decline in price, in part, has forced it to disgorge $100 billion from it’s balance-sheet.  The catalyst was a weak christmas quarter.  The quarter might have been saved if apple had new computers to ship for christmas, but they did not.  The 1 million unit shortfall in computers was about 1.2 billion in revenues.  If apple had paid even modest attention this part of the business, perhaps the quarter would have been saved.

Despite all the clever justifications, I think the real lesson is that Apple cannot walk and chew gum at the same time.  Their engineering and management abilities allow them to only focus on one thing at a time.

If management misunderstands the business they are in, its bad news.  They want us to think they are in the software business,  but apple if profoundly weak at software apart from OS and even the OS was purchased after years of failed efforts to develop a modern OS follow on to Mac 9

Apple is a platform company.  In their case that means an nice UNIX based OS running on decent hardware and IOS running on iphone.

However, does anyone think Lion and Mt Lion have been big improvements over snow leopard.  For every feature they gave they took something else useful away. The reliability and performance of the OS is worse not better than snow leopard.  Is IOS 6 and 7 competitive with or better than Android ? 

As it is now, I avoid apple OS updates.  There is not a single apple software product that I use (and not for want of trying).  I avoid icloud and who knows when they will abandon that initiative anyway.  I spend a fraction annually on apple products over what I spent 5 years ago and only because they dont have anything to sell that makes sense.

I’m one of those geek types in my small town who people come to ask advice about what to buy.  I used to point toward apple without question.  These days I dont.

Paul Goodwin

Too bad Apple didn’t spin off the Server business into a subsidiary focusing on enterprise. Had they done that, and staffed it to succeed, they probably would be selling not only the servers, but more Macs, iPads and iPhones along with them as integrated tool sets.


You mean Apple could have asked for $17K more just to handhold the users, and that would have resulted in a larger sale? Wow, imagine what their revenue would be then.


I love Snow Leo OS X on my 2009 MBP, hate Mountain Lion on my MB Air. Dislike their - Aapl - decision to drop 17” MBPs as I urgently would need a new one. I dislike using my iPad because of this dreaded spell checker I cannot disable. iPhone? Yes owning. No will not buy a new one as these devices are to Big for my pocket and the formerly gracefull and seamless integration to several services or borked. AppStore? Good for private ones but for business customers an impossition.

The future ? Definitively swinging back to Linux. Not only server wise. Will try to find notebook hardware that has good driver support by recent LTS Version of ubuntu.

My personell experience with Apple from previous four years: at the beginning a great concept, less troubles, nice to working with. Since OS X went into AppStore it went into pain. So the final conclusion: Marvelous machines provided by a company that does not care for customers view of Long Term planning. For any decision taker a big disappointment.


Good analysis, remarkable restraint grin

Apple only did well in the enterprise during the late 80s and very early 90s.  And even then, it was very hit and miss and a product owned by Apple one day could easily be spun off/sold/discontinued the next.  Think “Pathworks” or any of the products intended to help Macs integrate with mainframes.  Under Scully, Apple acquired a bunch of things, worked a little on them, made a little money on them and then ditched them.  All very disconcerting to us in the Enterprise trying to help the Mac proliferate.  We would make a little headway, then have the rug pulled out from under us.

By the time I joined in ‘93, they were already pulling back and when SJ returned, that door slammed shut.  The AIX server was a really good product, but it also ended up on the junk heap because Steve couldn’t stomach the licensing fees he had to pay IBM.  So once again, our customers were shafted.  I barely survived that debacle with a job to return to. 

Having the XServe product at all was a major coup, stemming from the impetus to sell into the science market where the customers were already familiar with UNIX and could do great things with such a product with very little hand-holding on Apple’s part.  But even that small investment was as much as SJ was ever willing to concede.  He NEVER like dealing with the Enterprise for precisely the reasons you stated. 

Perhaps the Mac Mini acting as a media server is the only kind of server that Apple can feel good about having in the product mix.  No one expects this product to have any sort of support ecosystem, since it is such a simple device.  And it is truly low enough cost that end customers buy it for their homes.  No messy enterprises to deal with.

Perhaps the life lesson here is the same one we should learn to adopt with individuals:  stop asking them (and Apple) to be something they are not and cannot be.  Our disappointments are ultimately a function of our expectations as much as, if not more than, of others’ failures.

John Martellaro

Lori: thanks for some great observations. (Lori and I worked at Apple together.)

Ed Ski

I’ve managed the Xserve and RAID for a small company. I must say, it was the easiest, sexiest server to setup, with storage and services, and management software and least expensive. (my boss would bring clients into the server room and show off the rack…so much for secure servers…). But it was not the easiest with MS integration (Active Directory, ...) .My first encounter with Apple’s arrogance was the “fixed” drive sled pricing. If you wanted to add capacity to the RAID, you were forced to buy Apple’s sleds with drives (Apple firmware for Xserve Admin to support). You couldn’t just get the sleds. And eventually (FW 1.5.1) there was support up to 750GB SATA drives. I managed to swap in other brand SATA drives, but I since I also had an iSeries IBM server and then some Dell servers (2800, 2900 series) to add in and that is where the CALs versus Apple’s unlimited clients were realized. But the Dell and IBM had the least drive losses per MTBF as they were all 5/yr SAS drives. I think those servers lost one HDD in 5 years (and under extended warranty). The Xserver lost both drives in less than 3 years, and the RAID lost 4 of the first 7 HDDs (Apple OEM) in 2 yrs of use. I eventually populated the 14 drives as RAID 5 with spares using non-Apple. But I moved on…
The Xserve and RAID were, an engineering visual work of art. Too bad the support was fleeting.
Then Apple EOL the RAID without atleast 6 months notice, and many of my Linux friends that used the RAID for storage were frustrated at Apple.
Like a previous commenter, I suspect they will do the same with the MacPro.

John Martellaro

Ed Ski.  You lost some of the drives in the Xserve and Xserve/RAID.  But did you lose any DATA?


An insider’s take always beats speculation however insightful! Good call. Even though I like DED he can be too brash and assertively arrogantly opinionated.

“As an aside, that inability to put technical and leadership support behind computational clusters led to Apple’s departure from supercomputing as well, but that’s another story for another day.”

I’m looking forward to it…

—I’m disappointed at the intellectually brilliant manpower Apple expends and then throws away wastefully!! I have said to S. Jobs that Apple could have been THE GREATEST COMPUTER COMPANY FOR EVER if it followed its innovative intellectual acumen. And Apple’s focus on the iOS will reduce its money and IP. S. Jobs himself should have kept his own advice and history and gone against the flow and direction –  its the only way that new directions are found and greatness is confered.

It is better to explode in a supernova rather than die as a brown dwarf or a red giant.


“….Certainly, Apple could have invested vigorous resources into securing an important market for itself in servers, storage and clusters.  ....”

The solution although with corporate pitfalls was to utilize its off shore billions. Instead of trying to repatriate profits Apple could have set up in Spain, Greece, Rumania or anywhere in EU a call center for Xserve/RAID corporate client support.* In 2008 Apple had about $20B at least in Europe. Apple could even have manufactured Xserve/RAID in Europe and repatriate its money hidden in the CPUs and hard drives back to Uncle Sam as a product! This action would maintain a product, keep Apple’s foot in the corporate door and clients happy as they paid for the service contract, as well as being a global manufacturing player and utilizing profits effectively to expand its brand.** Though that would make Stevee Baby an Indian God with many hands and heads—difficult for mere mortals and make him immortal!! 

IBM set up research centers to use its profits abroad.

The only down side would be accountant CEOs who could take Apple down IBM’s road!!

* I was amazed at the movement transcontinentally that USA workers made—corporate air travel expenses for a low level supervisor in shop fit out I guessed at greater than 60% of salary. Perhaps its the cheapness of air travel in USA that I’ve not appreciated.

** S. Jobs never saw any of his companies as conglomerates, they were all lean even today. Inventions, products or software not currently useful was ruthlessly excised, to Apple’s detriment later. This is a character fault of Stevee Baby from day one. (Instead of appeasing his partner Woz to make in-house the Basic Language Unit for Apple ][ he discarded his “inferior” co-founder and went to Bill Gates) The difference between Edison and Jobs is that Edison kept all of his babies because he knew his inventions were valuable – something Jobs never felt till iPods. Maybe because they were not his own inventions but “inferior” company droogs’.


I’ve just learned that Apple’s financial adviser is Goldman Sachs.

Ten years from 2011 and Apple is bankrupt!

Thanks to Tim Cook not heading S. Jobs hating of the Wall St. suits!

That’s what I think of Apple now and the destructive advice Goldman Sachs gives its clients!


“.....The only down side would be accountant CEOs who could take Apple down IBM’s road!! ....”

I never thought it would happen so soon: that Tim Cook could loose $140 billion in five years. What ever happens in eight years Goldman Sachs owns Apple and it has lost its reason to exist and will be just another Dell without its founder. 
Its America’s way LAND, CRAY, DELL, JOBS post mortum.
Good Bye American (Apple) Dream!!

Log in to comment (TMO, Twitter or Facebook) or Register for a TMO account