Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Much as I like the over-engineering on apple hardware, I don't think they had much of a chance to take their business model into the server world. Serving up FTP, Web, telnet, or some other IP based protocol is a pretty standard thing and there is lots of freeware out there that does the job on cheaply assembled hardware. None of the advantages of mac osx show in this role--the user interface and what all else is valued. There is no such thing as a 'better' HTTP server--its simple and either it works or not.

And with companies like facebook and google buying custom built servers by the tens of thousands, apples approach of a premium device simply can't compete. These companies simply write the specs for a server and let companies submit bids to build it, and the volumes are such they probably have their own special stripped down motherboards with minimal bells and whistles. They even get down to the savings level of only including the exact number of USB ports they want (if any) and certainly no use for things like onboard audio. Its a whole different dynamic. I think I read that some of these servers might not even have metal cases all the way around because they cost money and don't serve a purpose in these datacenters with acres of space under the roof.
I have to disagree with just about all of this. The companies I've worked for didn't buy bottom of the barrel, cheap systems to host their web sites. They paid for quality equipment such as HP, Dell, and Sun (Oracle) systems to do so. The Xserve was comparably priced with these systems and Apple was not at a disadvantage, price wise, with these systems.
 
I have to disagree with just about all of this. The companies I've worked for didn't buy bottom of the barrel, cheap systems to host their web sites. They paid for quality equipment such as HP, Dell, and Sun (Oracle) systems to do so. The Xserve was comparably priced with these systems and Apple was not at a disadvantage, price wise, with these systems.
Except that, unfortunately, Apple only had a single entry level 1U server model for sale. IO-constrained, and no scale-up options.

It was like a burger restaurant that served a single patty on a bun without any extras. Want ketchup? No. Want onions or a second patty? No. Mustard? No. Coleslaw on the side? No.

Why would you go there?

Most of my systems are ProLiant DL380 systems - 2U with 80 PCIe 3.0 lanes and six PCIe slots.

Quad Gb Ethernet on the mobo and dual 10 Gb Ethernet on mezzanine cards.

26 disk drives (52 TB) internal RAID drives with 4 GiB writeback cache.

Dual 1400 watt power supplies and support for dual Titan X or other dual-width cards.

Apple needed to either make a 2U "Xserve Pro", or kill the 1U Xserve. We know what they did.

(And I threw up in my mouth a little when I typed "Xserve Pro"...)
[doublepost=1475283850][/doublepost]
I have to disagree with just about all of this. The companies I've worked for didn't buy bottom of the barrel, cheap systems to host their web sites. They paid for quality equipment such as HP, Dell, and Sun (Oracle) systems to do so. The Xserve was comparably priced with these systems and Apple was not at a disadvantage, price wise, with these systems.
I agree completely with your disagreement.

Google, Microsoft Cloud, AWS and the others are special cases. They scale to tens to hundreds of thousands of servers per site, and have high availability infrastructure that deals with the situation that at any point in time 10% of the systems might be in a failed state. They're fine with a rack with 42 systems failing - the other 10,000 systems in the container will carry on. They buy custom or bottom of the barrel systems.

Companies that need a handful of servers go with mainstream systems with the reliability built into the hardware. (ECC, Mirrored RAM, RAID, ...)
 
Last edited:
Well, my employer has cisco UCS devices with each blade hosting about 100 virtual servers, thousands of virtual servers total, and I'm pretty sure it could beat up your employer... ;)

Well, data center convergence is fascinating to me and its where the action is right now. Multi-acre server farms are interesting as well, and limit the growth potential of traditional server manufacturers. Mea culpa, yes now that I think about it mac server failed in the marketplace long before mega data centers with custom designed servers. IMHO, it failed because they came late to the marketplace without an enterprise software solution--good hardware, but as has been pointed out, and I have seen it myself, some beautiful engineering in dell and other server products. Speculating, but that is what I recall--and I will leave it at that.
 
  • Like
Reactions: George Dawes
Well, my employer has cisco UCS devices with each blade hosting about 100 virtual servers, thousands of virtual servers total, and I'm pretty sure it could beat up your employer... ;)
Thousands of virtual servers? A drop in the bucket compared to my previous employer (who offers cloud based hosting as a component of their business) :) The lab environment hosts more than a thousand virtual servers.

With that said UCS is a nice setup.

Well, data center convergence is fascinating to me and its where the action is right now. Multi-acre server farms are interesting as well, and limit the growth potential of traditional server manufacturers. Mea culpa, yes now that I think about it mac server failed in the marketplace long before mega data centers with custom designed servers. IMHO, it failed because they came late to the marketplace without an enterprise software solution--good hardware, but as has been pointed out, and I have seen it myself, some beautiful engineering in dell and other server products. Speculating, but that is what I recall--and I will leave it at that.
It's my feeling the Xserve was never intended to be a data center class system but rather more of a computer closet based server. I have a friend who owns his own business and has such a computer closet configuration. An Xserve would work nicely in his environment. That is where I think the Xserve was targeted.
 
Except that, unfortunately, Apple only had a single entry level 1U server model for sale. IO-constrained, and no scale-up options.

It was like a burger restaurant that served a single patty on a bun without any extras. Want ketchup? No. Want onions or a second patty? No. Mustard? No. Coleslaw on the side? No.

Why would you go there?

Most of my systems are ProLiant DL380 systems - 2U with 80 PCIe 3.0 lanes and six PCIe slots.

Quad Gb Ethernet on the mobo and dual 10 Gb Ethernet on mezzanine cards.

26 disk drives (52 TB) internal RAID drives with 4 GiB writeback cache.

Dual 1400 watt power supplies and support for dual Titan X or other dual-width cards.

Apple needed to either make a 2U "Xserve Pro", or kill the 1U Xserve. We know what they did.

(And I threw up in my mouth a little when I typed "Xserve Pro"...)
[doublepost=1475283850][/doublepost]
I agree completely with your disagreement.

Google, Microsoft Cloud, AWS and the others are special cases. They scale to tens to hundreds of thousands of servers per site, and have high availability infrastructure that deals with the situation that at any point in time 10% of the systems might be in a failed state. They're fine with a rack with 42 systems failing - the other 10,000 systems in the container will carry on. They buy custom or bottom of the barrel systems.

Companies that need a handful of servers go with mainstream systems with the reliability built into the hardware. (ECC, Mirrored RAM, RAID, ...)
they had the Xserve RAID
 
I don't know about discontinued, but there are hints of design and manufacturing challenges:


What the hell is going on in Apple's US-based Mac Pro factory?

http://www.theverge.com/2016/12/21/14037030/apple-made-in-america-failure

Three years on, the Mac Pro is ripe for an upgrade with its chips and connector ports lagging rival products. Because of the earlier challenges, some Apple engineers have raised the possibility of moving production back to Asia, where it's cheaper and manufacturers have the required skills for ambitious products, according to a person familiar with those internal discussions.​
 
  • Like
Reactions: Synchro3
The status of US manufacturing in general and Apple's partner plant in Austin is part of the equation. That said, which scenario do those facts support:
1) Discontinuation of the MacPro
2) Moving manufacture back to Asia
3) Building the nMP 7,1 in the USA to create political good will

Depending on the angle you choose, I think you could support all of the above from the same data points. I would add that the 3 year+ holding pattern can be viewed through different lenses as well.

Since I would love to see new computer hardware be more efficient and quieter, I appreciate the concept of the nMP form factor. I'd like to think that Apple expected to have die shrunk parts available by now that would have rendered the size and TDP limitations of the small cylinder moot. IMO, they overshot their target. If they had made the cylinder tall enough to host standard PCIe cards, put in a 650W PSU and offered a super-cooler pad for it to sit on for heavy duty use cases they might have had a winner. Sure, a lot of us need more than 2 internal PCIe slots, etc - but the percentage of use cases that needed more total grunt than the "tall can" form factor could support would have been substantially less than the cohort of underserved pros defecting to Windows/Linux from the 6,1.

Call me wildly optimistic if you want, but here's what I'm hoping:
Apple waits for die shrunk chips that can fit into the TDP envelope of the current cylinder design, then releases the best performance per watt workstation on the market. Iterative improvements in cooling design make thermal throttling of the GPUs rarely necessary. Apple gets serious about supporting eGPU over TB3, including proper drivers for nVidia GPUs. This actually happens before my cMP dies or gets left behind :)
 
  • Like
Reactions: JamesPDX
The status of US manufacturing in general and Apple's partner plant in Austin is part of the equation. That said, which scenario do those facts support:
1) Discontinuation of the MacPro
2) Moving manufacture back to Asia
3) Building the nMP 7,1 in the USA to create political good will

Depending on the angle you choose, I think you could support all of the above from the same data points. I would add that the 3 year+ holding pattern can be viewed through different lenses as well.

Since I would love to see new computer hardware be more efficient and quieter, I appreciate the concept of the nMP form factor. I'd like to think that Apple expected to have die shrunk parts available by now that would have rendered the size and TDP limitations of the small cylinder moot. IMO, they overshot their target. If they had made the cylinder tall enough to host standard PCIe cards, put in a 650W PSU and offered a super-cooler pad for it to sit on for heavy duty use cases they might have had a winner. Sure, a lot of us need more than 2 internal PCIe slots, etc - but the percentage of use cases that needed more total grunt than the "tall can" form factor could support would have been substantially less than the cohort of underserved pros defecting to Windows/Linux from the 6,1.

Call me wildly optimistic if you want, but here's what I'm hoping:
Apple waits for die shrunk chips that can fit into the TDP envelope of the current cylinder design, then releases the best performance per watt workstation on the market. Iterative improvements in cooling design make thermal throttling of the GPUs rarely necessary. Apple gets serious about supporting eGPU over TB3, including proper drivers for nVidia GPUs. This actually happens before my cMP dies or gets left behind :)
TB3 is still only max pci-e X4 3.0 and that limits video cards.
 
Call me wildly optimistic if you want, but here's what I'm hoping:
Apple waits for die shrunk chips that can fit into the TDP envelope of the current cylinder design, then releases the best performance per watt workstation on the market. Iterative improvements in cooling design make thermal throttling of the GPUs rarely necessary. Apple gets serious about supporting eGPU over TB3, including proper drivers for nVidia GPUs. This actually happens before my cMP dies or gets left behind :)
There may be an alternate universe where this is true. Unfortunately we live in this one. :(
 
TB3 is still only max pci-e X4 3.0 and that limits video cards.
Not suggesting that any eGPU solution will harvest the full performance potential of the card. That said, some computational operations are complex enough that the size of the data pipe presents a minimal bottleneck. More to the point, we need robust driver optimization for the latest cards - starting with nVidia Pascal - that will require the kind of engineering resources deployed with genuine cooperation between Apple and nVidia that clearly doesn't exist.

I'd like to think that companies who put their egos in front of improving their products/services would be weeded out by market forces... but this is a tech forum, not a political/economic theory salon. That said, I could argue that the best chance we have of getting a truly badass workstation from Apple (with proper support for all leading GPU vendors) is if Tim Cook and the leadership get their egos in the game. Perhaps we could heckle them into proving they really can make the next generation powerhouse that makes Windows PCs cower. I still remember the joke Intel logo featuring a snail icon back when Macs ran circles around most de-contented PC clones.

Hey, you wussies running what used to be a computer company, how does it feel to have ASUS eating your lunch.

Not only can you not innovate your own ass, I don't think you could find it to check.
 
  • Like
Reactions: JamesPDX
Not suggesting that any eGPU solution will harvest the full performance potential of the card. That said, some computational operations are complex enough that the size of the data pipe presents a minimal bottleneck. More to the point, we need robust driver optimization for the latest cards - starting with nVidia Pascal - that will require the kind of engineering resources deployed with genuine cooperation between Apple and nVidia that clearly doesn't exist.

I'd like to think that companies who put their egos in front of improving their products/services would be weeded out by market forces... but this is a tech forum, not a political/economic theory salon. That said, I could argue that the best chance we have of getting a truly badass workstation from Apple (with proper support for all leading GPU vendors) is if Tim Cook and the leadership get their egos in the game. Perhaps we could heckle them into proving they really can make the next generation powerhouse that makes Windows PCs cower. I still remember the joke Intel logo featuring a snail icon back when Macs ran circles around most de-contented PC clones.

Hey, you wussies running what used to be a computer company, how does it feel to have ASUS eating your lunch.

Not only can you not innovate your own ass, I don't think you could find it to check.
The pathetic thing about Apple is that ever since Jobs was at the helm of the company, they have garnered a sense of self-fellating grandeur of which the customer is NEVER right. And while I can sometimes agree with that statement, in the case of Apple in its current time, they stick to that statement like goddam gospel.

They'll never admit they're wrong but rather keep self-fellating themselves dry.

I've met some people who have worked for Apple and they're some of the smuggiest pieces of **** I've ever encountered... not an ounce of humbleness to them. The philosophy of a superiority complex is rampant among those who work at that company.
 
When Phil presented the nMP on stage at WWDC he was quoted as saying "a new Mac Pro for the next ten years". We've still got a long way to go before ten years is up. The Mac Pro isn't dead. Just neglected.
 
6 more years.
It's time to face the obvious:

hes-dead-jim[1].jpg

Macintosh - b 1984, d 2013
 
Last edited:
When Phil presented the nMP on stage at WWDC he was quoted as saying "a new Mac Pro for the next ten years". We've still got a long way to go before ten years is up. The Mac Pro isn't dead. Just neglected.
Sadly, he meant the exact same model, it appears.

I wonder if Phil's good for another six?
 
I just bought something at an Apple Store and went to pick it up. I spoke with a salesperson who said she's talked to a lot of people who were pretty pissed off about the lack of Mac Pro updates. Then she tried to sell me an iPad.

After the purchase I got an email from Apple asking me to take a survey about my experience. After rating the salesperson the survey went on to ask more than a few questions about whether the salesperson told me about the iPad, iPhone, etc. It also asked if I knew about iPhone trade in (or related?) program and so on.

It was ALL about iOS devices. Not even a single question about desktop computers.

RIP Mac Pro.
 
I just bought something at an Apple Store and went to pick it up. I spoke with a salesperson who said she's talked to a lot of people who were pretty pissed off about the lack of Mac Pro updates. Then she tried to sell me an iPad.

After the purchase I got an email from Apple asking me to take a survey about my experience. After rating the salesperson the survey went on to ask more than a few questions about whether the salesperson told me about the iPad, iPhone, etc. It also asked if I knew about iPhone trade in (or related?) program and so on.

It was ALL about iOS devices. Not even a single question about desktop computers.

RIP Mac Pro.

This is part of the key though -- Apple doesn't gather a lot of feedback from informal sources ( most companies do not ), if you're not emailing Tim or filling out these surveys they assume the world is grand unless sales drop.

One would think that the nMP sales have plummeted in the past year though as I think people thought at least a bump on the graphics cards or something was going to happen.

Here's hoping we hear something at the next media event -- even if they say it's been discontinued at least it stops people waiting and hoping and they can begin the move to Windows for their workloads
 
I don't know about discontinued, but there are hints of design and manufacturing challenges:


What the hell is going on in Apple's US-based Mac Pro factory?

http://www.theverge.com/2016/12/21/14037030/apple-made-in-america-failure

Three years on, the Mac Pro is ripe for an upgrade with its chips and connector ports lagging rival products. Because of the earlier challenges, some Apple engineers have raised the possibility of moving production back to Asia, where it's cheaper and manufacturers have the required skills for ambitious products, according to a person familiar with those internal discussions.​
Seems more like over engineering was the problem. Over-engineering is bad engineering. Apple too frequently throws in finicky quirks to look 'cool' that merely mean higher costs and manufacturing issues.

A MacPro just needs to be a big fat tower. Sure, make the exterior pretty, make it rose gold for all I care, but fundamentally it doesn't have to differ from any other workstation on the market.

Apple's trump card was Mac OS X. The Ive designs should have been the cherry on the cake but they became the cake. Mac OS X was the prefect enterprise OS and they should have made some big UNIX acquisitions. Even if that didn't bring in the big money they would have been much better placed for the Cloud era. They could have still wowed the consumer market.

Ultimately Tim Cook is a bean counter. If there's more beans in consumer junk that's where he'll put the energy, even if in some cases low-bean activities are important for the bigger picture. Of course, Apple are still harvesting money with ease right now but you only have to look at Nokia to see how a top dog can become a complete nothing in the consumer space in the blink of an eye.
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.