In a bit of a twist where many companies are focused on AI systems and Intel Xeon 6 launches, HPE is launching a new series of 4th Gen Intel Xeon Sapphire Rapids-based servers. These are not just any servers, Instead, these are the HPE NonStop Compute X5 generation designed for higher-reliability x86 compute.
HPE NonStop Systems Get an Intel Sapphire Rapids Refresh
The new systems come in two flavors. First is the Intel Xeon Bronze-based NS5 X5. This is the entry configuration that scales to up to four sockets with up to 512GB of RAM and 20 networking ports per system.
The second one is the HPE NonStop Compute NS9 X5. This can scale up to 16 Intel Xeon Gold 6400 CPUs and 8TB of memory with up to 270 networking ports per system. Part of the value of this configuration is, of course, that the scalability is higher than the more mainstream 4-socket and more niche 8-socket general purpose Xeon servers.
Still, at this point you might be wondering why in 2025 HPE is launching new systems with 2+ year old, and two generation old processors. The idea behind these systems is to be at minimum five 9’s or six 9’s (99.999% to 99.9999%) reliability. Or to put this segment in some more context, the new IBM z17 is designed for eight 9’s or 99.999999% uptime (or 1s of downtime per year.) We showed the lengths that IBM has to go through to get to that figure all the way from the liquid cooling down to analyzing individual silicon atoms. You can learn more about that in our popular video:
At the same time, the IBM Z is not widely known as being low-cost. As such, HPE NonStop is designed to fill a void where lower cost than traditional mainframes is required, but there are also requirements for higher reliability and scale-up than standard servers.
Final Words
This is a line that we do not hear about often, but it is neat. We thought given some of our recent mainframe coverage that we should balance it out with this HPE segment. Hopefully we will get to see these at HPE Discover 2025 in a few weeks.
Good to hear that HPE is still maintaining the good old Tandem-initiated NonStop system. I guess there are still businesses running on HP’s own HP 3000 mainframe? DEC-initiated OpenVMS, and of course Oracle Solaris, IBM Power System, IBM i (OS/400…), and more?
Those proprietary systems used to be at the forefront of hardware innovation when Intel was the follower, and now it flips around and those proprietary systems are often one or two generations behind the Intel ecosystem. IBM might be the exception though.
@CheKim
Yup, IBM is really the only one left in the hardware game from the highest end enterprise vendors, and doing quite well.
HP abandoned proprietary processor designs with the Itanium debacle, DEC Alpha is long gone with OpenVMS going native x86, Oracle/Fujitsu SPARC are also abandoned since 2017. Fujitsu moved to ARM, and Oracle is pretty comfortable with x86.
HPE even ported the NS/OS to VMWare. But its not as flexible as one would think. While this is essentially a hardware refresh, HPE doesn’t even provide field support for their Non Stop platforms anymore. It’s been farmed out to a third party.
This brings all the benefits of a consolidated platform, but it also carries some of the risks as well.