High Performance Compute

High Performance Compute is designed for large scale workloads (e.g. Grid Engines) that have a long-term requirement for cost-effective and highly scalable compute resources. The service leverages UKCloud’s trusted  cloud platform that is connected to all the key government networks.

The inherent scale and elasticity of cloud makes it an ideal environment for large-scale compute-intensive workloads such as Grid Engines, High Performance/Throughput Compute (HPC/HTC) and Massively Parallel Processing (MPP), some of which has an enduring requirement for large amounts of processor cores and memory.

UKCloud’s High Performance Compute service is designed for these types of workloads. In return for a volume and term commitment for a large number of compute cores, customers get a cost-effective, highly scalable infrastructure to support their compute intensive solution.

Features & benefits

There are many features and benefits to adopting High Performance Compute, including being highly scalable for throughput computing and optimised for massively parallel processing (MPP) and grid engines.
Find out more

Service options

We provide recommendations on the best service options for you based on three workload characteristics; workloads processing sensitive data sets, workloads requiring connectivity to lower security networks and workloads that require high core-to-core memory ratio.
Find out more

Technical specifications

High Performance Compute includes a number of technical specifications as standard including virtual firewalls, basic load balancing and DDoS protected internet.
Find out more

Service information

The service is provided with a variety of cluster-wide options. View our full service information to gain insight on the options offered.
Find out more

There are many features and benefits to adopting High Performance Compute, including being highly scalable for throughput computing and optimised for massively parallel processing (MPP) and grid engines.

Feature Benefit
High performance, massively scalable for high throughput computing A solution that can meet the largest of requirements to de-risk your project
Optimised for massively parallel processing (MPP) and grid engines Superior value from a purpose – built solution
Choice of VM/node sizes and configurations Flexibility to match the requirements of your workload
Extensive independent testing and validation of assurance A secure and trusted platform which addresses risks associated with hyperscale providers
Connected to PSN Assured, PSN Protect, N3/HSCN, Janet and Internet Easy interconnection to existing systems and accessible by your current users and devices
Your workload characteristics UKCloud recommends
Security domain Core-to-memory ratio Connected networks
Workloads processing particularly sensitive data sets Elevated OFFICIAL 1core:12GB HybridConnect
VM and associated data reside in the non-internet-connected Elevated OFFICIAL security domain, providing the highest levels of trust and assurance. VMs have an uncontended compute (CPU/GB) resource allocation.
Automated rebalancing is disabled to reduce workload movement around the platform, reducing workload disruption.
Connected to higher-security networks such as PSN Protected, legacy government networks (such as GSI, PNN and CJX) and RLI
Workloads requiring connectivity to the internet and other lower security networks Assured OFFICIAL 1core:12GB JAanet
VM and associated data reside in the internet-connected Assured OFFICIAL security domain, providing access and connectivity to lower-security networks. VMs have an uncontended compute (CPU/GB) resource allocation.
Automated rebalancing is disabled to reduce workload movement around the platform, reducing workload disruption.
Connected to lower-security networks such as PSN Assured, Janet, N3/HSCN and DDoS-protected-internet.
Workloads that require an especially high core-to-memory ratio Assured OFFICIAL 1core:16GB DDoS-protected internet
VM and associated data reside in the internet-connected Assured OFFICIAL security domain, providing access and connectivity to lower-security networks. VMs have an uncontended compute (CPU/GB) resource allocation.
Automated rebalancing is disabled to reduce workload movement around the platform, reducing workload disruption.
Connected to lower-security networks such as PSN Assured, Janet, N3/HSCN and DDoS protected Internet

High Performance Compute includes as standard:

  • Virtual firewall
  • One production virtual data centre (VDC) and one non-production VDC
  • VMware HA (High Availability) protection (only with persistent storage)
  • Basic load balancing
  • DDoS-protected internet

The service is provided with a variety of cluster-wide options as follows:

  • Choice of our natively internet-connected Assured OFFICIAL domain or our Elevated OFFICIAL domain which is primarily connected to government secure networks.
  • Choice of network connections including DDoS-protected Internet, PSN Assured, PSN Protected (including legacy networks such as GSI, PNN, CJX), RLI, N3/HSCN and Janet.
  • Choice of virtual CPU mode. Compute Intensive Infrastructure is based on hyper-threaded cores which provide a cost-effective solution for compute-intensive workloads. You can opt for Native mode by using up to half the allocated cores to ensure a one-to-one mapping between virtual CPUs and physical (non hyper-threaded) cores for maximum processing power.
  • Choice of core-to-memory ratio. All VMs in your compute-intensive cluster must be identically configured based on one of two core-to-memory ratios.