This Document has limited distribution rights. It may not be distributed to the general public or shared beyond the employees of the licensee company. Upon the termination of an employee, whether voluntary or involuntary, access to this document must immediately cease. If the licensee company wishes to utilize this information in outward facing communications, please contact firstname.lastname@example.org.
On May 16, 2017, Nutanix announced that it had formed a partnership with IBM to deliver Nutanix’s hyperconverged software using IBM’s POWER processor. So naturally, the question begs to be asked, is hyperconverged infrastructure (HCI) defined by the use of x86/x64 CPUs? Neuralytix does not define HCI by the processor type, so the answer to whether a POWER-powered cluster of nodes using Nutanix software is technically HCI.
But taking the “technicality” out of the equation, what does a POWER driven HCI cluster mean? Well, putting it simply, it becomes a highly dense, highly powerful Linux cluster capable of delivering many more virtual machines (VMs) than a traditional x86/x64 cluster, and also capable of more network bandwidth. It is Neuralytix’s opinion, that the differences in storage capabilities are nominal, since it is possible to create quite large storage pools with x86/x64 processors (essentially, the limitation is on the size and number of hard disk drives [HDDs] and solid state disks [SSDs]).
Neuralytix also posits that it is analogous to Windows NT deployed on Itanium chips. Difference underlying architecture, but still capable of running (adapted) Windows applications. There is one exception, in that, the Linux operating system, and the KVM hypervisor can conceivably abstract the architecture, and run x86 workloads on the POWER platform (again, we stress conceivably).
The POWER driven HCI cluster is akin to the Digital Equipment Corporation’s (DEC’s) VAX and MicroVAX machines. While the MicroVAXes delivered a lower amount of compute capability, the larger VAXes were capable of addressing a larger memory pool and a higher number of more powerful processors.
This means that demanding scale-up workloads, such as SAP or Oracle, can now run more effectively on the POWER based platform. It also means that larger in-memory databases can be stood up.
All of these analogies brings us back to a basic direction – hyperconvergence (and ultimately composable infrastructure) will allow us to build the 21st century mainframe. What the POWER architecture gives, is that it extends the core/processor and memory densities beyond what the x86/x64 architecture currently provides. The latest Intel E7 24-core processors can go up to 8 sockets inside a server, delivering 192 cores at 2.4Ghz base frequency; while the largest IBM Power System E880 server can deliver a maximum of 192 cores also, but a base frequency of 4.02GHz.
This is all a matter of leapfrogging technology. IBM’s POWER9 chip is expected to deliver 24 cores per processor (doubling the current POWER8 generation) also at around 4GHz, so one should expect that the largest IBM POWER-powered system will deliver 384 cores at 4GHz.
With many new applications being written for containers, the number of cores could make a significant difference in the future. For now, Neuralytix’s opinion is that the POWER-powered HCI architecture is interesting for a small few, but in no way threatening to the status quo.