ReliaCloud Hosted Private Cloud

ReliaCloud Edge Nodes

ReliaCloud Hosted Private Cloud

High Performance Compute

ReliaCloud Edge High Performance Compute (HPC) is configured to support high performance workloads, such as I/O intense applications, large high-performance databases, or volumes based storage services.

To support such workloads and maintain a high level of performance, we’ve outfitted the HPC node with NVMe storage.

ReliaCloud Edge HPC Nodes

Model ReliaCloud Edge HPC.3661N

ReliaCloud Edge
HPC.3661N 

ReliaCloud Edge HPC.3661N

ReliaCloud Edge
HPC.4830N 

Price Per Node* $3975 AHV
$4175 ESXi
$4285 AHV
$4545 ESXi
Chassis/Model Nutanix NX-G8 8070  Nutanix NX-G8 8070 
Processor 2x Intel Xeon-Gold 6354 Processor (3GHz/ 18-core/ 205W) (36 cores)  2x Intel Xeon-Gold 6342 Processor (2.8GHz/ 24-core/ 230W) (48 cores) 
RAM** 1024GB  512GB 
SSD 61.44TBNVMe  30.72TBNVMe 
HDD None None
NIC 2x 25GbE, 2-port, SFP28 Network Adapter  2x 25/10GbE, 2-port, SFP28 Network Adapter 
GPU None None

*VMware ESXi licensing not included. OneNeck’s VMware Cloud Provider Program Licensing is required.

**Typical configuration for 1024GB RAM is 32x32GB 3200MHz DDR4 RDIMM.

 

Other Details

Model HPC.3661N  HPC.4830N 
Licensing Capacity Based Capacity Based
Cluster Software AOS ULT + Prism ULT  AOS ULT + Prism ULT 
Min Nodes per Cluster  3 3
MAX Nodes per Cluster 12 12

Sizing Guidance

Model HPC.3661N  HPC.4830N 
OneNeck Benchmark Score*  149 60
VMs per Node  50  20
Largest Size VM – vCPU  18 24

*OneNeck Benchmark Score – A VM average size is used as the baseline for setting a standard configuration to track price points across cloud solutions. This includes 2vCPU, 8 GB of RAM and 300 GB of storage.

The VM sizing was calculated across OneNeck’s study of our ReliaCloud customer base, and we’ve taken this VM sizing to give an approximate indicator of how many benchmark VMs we could safely fit into a 3-node cluster comprised of a specific node type.  Settings assume the follow parameters:

  • Per VM: 210 GiB cold and 90 Gib hot data
  • Cluster:  AHV, 4 to 1 vCPU : Physical core ratio, RF2 data resiliency, N+1 fault tolerance to 85% resource thresholds, in cluster scale out Prism Central and local data protection with snapshot schedule 4xhourly,7xdaily, 4xweekly, 1xmonthly.