What I never understood...
Is that these systems are designed to have stripped back features, removing a lot of HPE Proprietary technologies and software. The default warranty is also parts-only on these machines, which is quite unique in the server world where at least 1 site of labour and on-site cover is standard on pretty much everything. The intention behind all this is to reduce the cost and complexity of each system, making them cost effective to deploy at scale and rely on whatever tools said company has developed or bought in to manage all the flashy scale out cloudy stuff.
If I were being cynical I would sum that up as "putting a HPE badge on a white box Foxconn server", but in fairness I'm not 100% sure that's what they actually did.
But as far as I can see they cost more than ProLiants on a like-for-like basis, as well as being less flexible in configuration and expandability.
So what's the effing point?
Re: So what's the effing point?
you can say that about HPE these days.
Re: What I never understood...
These servers have a different use case. They are deliberately feature stripped to save power. So far as being more expensive...its an opex discussion rather than an acquisition cost discussion. Also, you have to price things higher In order to show bigger discounts to Microsoft- they get a better discount than a normal shop.
These servers are not aimed at small shops with less than hundreds of identical servers.
Re: What I never understood...
I worked in Dell server support over 10 years ago and we had plenty of customers that were parts only or had parts onsite they replaced themselves and called up to replace stock. Not the norm but did exist.
I wish Mystic Meg would just cut to the chase. She doesn't want to make anything or design anything or even deal with the messy business of "customers" ... she cares about nothing besides quarterly earnings.
She should sell everything, invest the resulting cash pile in Bonds/Gilts/Treasury Securities (maybe a real estate portfolio) and stop wasting everyone's time pretending to be a technology company.
How could they possibly compete against ODMs?
in Dell's case, they flipped over to using ODMs years ago and largely stopped their own manufacturing. HP finally figured out how the DellDCS and Hyperscale teams worked way too late and attempted to mimic them even as Dell was exiting that phase of the business ...
At some point, x64 is just x64 and no one cares about the logo on the bezel -they don't care if there even is a bezel. They do care about density and power and thereby cooling efficiency once you start getting to cloud scale. Dell has had more power efficient server configurations than HP for over a decade. You don't need to be *much* more efficient at this scale, as a single watt per server saved translates to kilowatts at this scale. Just slightly more efficient is enough to tilt Opex savings your way.
Then a lot of the smart Dell guys behind that initiative got hired away to Facebook and the like to teach those former customers how to just go direct and build even better kit.
The toothpaste is out of the tube now...buying cloud scale compute from HPE costs more money for the customer while not generating enough margin to pay for the effort, particularly considering the layers of PHBs at HPE that must get paid for the effort.
If the channel chief quoted is actually telling the truth, I predict that discreet component choices will become more limited by HPE in order to protect margins. The result will be fewer sales as the components chosen will not suit as many use cases and better options exist by going straight to the ODM. HPE moves too slowly to keep up in this space. This business unit is circling the drain.
It's not often that I will say it, but Meg is right here...better to just shut it down before losing your one anchor customer does it for you by going direct.
I can think of at least one reason
Interestingly I think I might be able to add something around this. I was looking at Tech Field Day on YouTube and found an interesting article around the Open19 Project. Apparently LinkedIn have been using something similar for some time now and now that Microsoft have acquired them I would assume they would go down this path rather than buying into something completely white box. That and the fact that things like OCP and Microsoft's own node designs have been in research for some time now I don't know why you would go ahead and buy something standard from a major vendor.
Check it out: