Data Science Infrastructure HSS TOP

# Please purchase the extended shared storage and pool type computing resources from FY2023 to FY2024 during in grace period(April 2 to April 16).

The Shared Use Computers HOKUSAI consists of the Supercomputer HOKUSAI BigWaterfall (HBW) and the Data Science Infrastructure HOKUSAI SailingShip (HSS).
HSS provide system infrastructure to users as tenants by virtualizing computing resources/shared storage/networks.

1. Data Science Infrastructure HOKUSAI SailingShip(HSS) system

System Overview
  • Operation schedule: From October 19, 2020 to May, 2026
  • Data farm
    • 30 PB、EXAScaler (Lustre) filesystem
  • CPU farm
    • FUJITSU Server PRIMERGY CX2550 M5 440 node、21,120 core、1.62 PFlops
    • Intel Xeon Platinum 8260 (2.40 GHz、24 core)、2 CPU/node
    • 384 GB (DDR4-2933)、SSD 1.92 TB、10GBASE-Tx2
  • PIaaS (Private Infrastructure as a Service)
    • Mirantis Cloud Platform (MCP): Commercial distribution of OpenStack
    • CPU farm, data farm, and network are virtualized by MCP, and system infrastructure is provided as a tenant for each user.

The data farm can also be used as storage infrastructure for the RIKEN Research Data & copyrighted work Management System (only from within RIKEN).
As with normal use of SS, you can use the shared storage reserved after project creation.
However, although each usage method shares the same area, they cannot access each other’s data.

If you are interested in using a GPU for AI-related research, please refer to RAIDEN, a large-scale computer for deep learning analysis (only from within RIKEN) owned by the the RIKEN Center for Advanced Intelligence Project.

2. Usage Fees

The services and usage fees that are subject to usage fees are defined in the Handling of Usage Fees for Information System Services Provided by the RIKEN Information R&D and Strategy Headquarters.

Pool type CPU resources
  • Purchased in advance and monthly basis until the end of the fiscal year
    • No prorating, even in the middle of the month
  • 360 JPY per month for 2vCPU
    • Allocate 8GB of memory and 36GB of local storage (SSD) per 2vCPU
On-demand type CPU resources
  • Purchase resources in advance, and only the amount you actually use will be deducted from them
    • Please note that there are many instance states. For details on which state is subject to billing, please refer to the HOKUSAI SailingShip User Guide “3.12 Instance State after Startup”.
  • 360 JPY per 720 hours for 2vCPU (1CPU core)
    • Allocate 8GB of memory and 36GB of local storage (SSD) per 2vCPU
Shared storage
  • Purchased in advance and monthly basis until the end of the fiscal year
    • No prorating, even in the middle of the month
  • 180 JPY per month per 1TB
    • 270,000 inodes are allocated per 1TB.

You can purchase points through the Online Submission System and use the points to reserve computing resources and storage for the amount of the purchase.
The unit of purchase for points is 360 points, which costs 360 JPY.
The points are valid until the end of the fiscal year and cannot be returned.
Also, once the points are converted to service content, they cannot be converted back to points.

Expenditure Transfer Schedule

The payment of user fees will be made through expenditure transfers. The expenditure transfer will be made according to the following schedule.
For example, if you purchase 30,000 yen worth of points in April, the full amount will be expenditure transferred in July.

  • Purchased by the end of June: Expenditure transfer in July
  • Purchased by the end of September: Expenditure transfer in October
  • Purchased by the end of December: Expenditure transfer in January
  • Purchased by the middle of February: Expenditure transfer in February
  • Purchased by the end of the fiscal year: Expenditure transfer in March

Applications approved by the deadline will be processed. If you need special handling, please contact us by email.

Notes on external funds

For MEXT (excluding KAKENHI, JSPS, and JST), JAXA, NEDO, METI, and MLIT proposals, please contact the External Funds Office to inquire about the availability of payment, payment methods, deadlines at the end of the fiscal year, and other conditions.
[Contact] gaibushikin-toiawase [at]
If you are using external funds, please make sure you use up your points, computational resources and storage.

3. Usage Agreement and Operation Procedure

Please read and observe the Usage Agreement and Operation Procedures before use.

Supercomputer Systems Usage Agreement Enforced: April 1, 2024
Supercomputer Systems Operation Procedure Created: April, 2024

For more information on how to use and apply for HSS, please refer to the Online Submission System (hss-desk)
to download the HOKUSAI SailingShip User’s Guide and Online Submission System User’s Guide.

4. SS usage guide

How to Apply for a Project
  1. Create a project in the Online Submission System.
  2. Purchase points in the Online Submission System and use the points to secure computational resources and storage.
  3. Use the OpenStack management console to build a tenant (system infrastructure) in a virtual environment.
  4. Using Virtual Machines (VM) and shared storage within the tenant.
Trial Use

Trial use is available for first-time users of SS.

Support for PIaaS usage in SS

We will provide support for building a system on the tenant as PIaaS usage support.

Project Management and Security Export Control
  • Since HSS is like a tenant (system infrastructure) rent, the payment representative must have control over what goes on the project as with the laboratory computer servers.
  • If any of the project members or tenant users are foreign nationals or non-residents who have not been approved on screening forms by the Research Integrity and Economic Security Division (see Security Export Control Policies), they are required to apply to the Research Integrity and Economic Security Division for screening forms. In such cases, the project member should submit the application.
  • RIKEN requires approval of the screening form when hiring or accepting foreign nationals and non-residents. Those whose screening form has been approved do not need to apply again.
Default resource limits and change requests

The following limits are set by default for point purchases and computing resources.

  • The maximum value for point purchases is 1 million JPY.
  • The maximum value for shared storage defaults to 100TB.
  • Maximum number of execution vCPU for on-demand type is 768 vCPU.

You can add four different network addresses for your project by email .
Please fill in the project name, the resource to be changed, the value to be changed, and the purpose of use.

Addition of network addresses

You can add four different network addresses for your project by email .
(See the HOKUSAI SailingShip User’s Guide for details on network addresses.)
Please fill in the project name, the type of network address, the number of IP addresses required, and the purpose of the network address.

When you request the addition of a RIKEN global FIP address, please follow the steps below, although details will be provided each time.

  1. Request the addition of a RIKEN global FIP address here.
  2. Configure the VM with the assigned RIKEN global FIP address.
  3. Apply for a global address from the Service Portal for auditing and DNS settings.

For security reasons, we will remove the RIKEN local FIP from the project after the RIKEN global FIP setting is completed.

5. Online Submission System (hss-desk)

  • Access is via the RIKEN authentication infrastructure
  • Project Application and Approval
    • Apply for new projects and manage project information and members
    • Purchase points and resources
    • View and approve applications
  • Guide
    • HOKUSAI SailingShip User’s Guide
    • Online Submission System User’s Guide
    • Point simulator

6. OpenStack Control Console (hssa cluster, pool type) (hssb cluster, on-demand type)

  • Accessible from inside RIKEN (or RIKEN VPN).
  • Tenant management: Building VMs, setting up networks, etc.


Please be sure to include the project name when inquiring about a project in use.