Frequently Asked Questions
- What is changing?
- Why the change?
- What are the differences?
- What are the network latency implication between Deakin and NCI?
- What are the implications of data transfers between RDSI node and raijin?
- How will I obtain access?
- I am an existing VPAC user, will I be automatically migrated?
- How large a capacity has Deakin acquired?
- What are the Deakin University NCI Resource Allocation Committee terms of reference
1. What has changed?
Deakin has changed its peak HPC facility it has discretionary access to from the Trifid system (which was managed by VPAC and has been decomissioned) to the Raijin system (managed by NCI)
2. Why the change?
Deakin has become a member of NCI to leverage the capacity of the NCI peak facility. This will help Deakin researchers prepare successful merit allocation proposals (i.e. NCMAS) in the future.
The NCI Raijin system has more capacity, is cheaper to use and is strategically aligned to other Australian research platforms. As well as the improved capacity, the support provided by NCI staff, which will ensure that researchers are able to get the most out of the facility.
3. What are the differences?
- Trifid is physically located in Victoria (at an RMIT premises)
- Trifid specifications (no longer available)
- 10Gbps dedicated research network connectivity
- Raijin is located in Canberra (at ANU)
- Raijin specifications
- 1Gbps shared network connectivity (to be upgraded to 10Gbps in Q2 2014)
- Alignment with NeCTAR Research Cloud at ANU and RDSI node
- NCI will bring an enhanced delivery in both capacity computing (i.e. number of large jobs) and capability computing (i.e. enabling massive highly parallel jobs)
4. What are the network latency implication between Deakin and NCI?
Any bottle necks will be at the Deakin end. It is something for the researcher to consider (or recognise) if they intend to move large data files around. If you are transferring data from your desktop/laptop and are connected to a network point in the wall, you will be using a 1Gbps connection. If you are moving data from a server in the data centre, then that will move over the WAN at its full capacity - currently that is a 10Gbps conection to AARNet.
5. What are the implications of data transfers between RDSI node and raijin?
Latency of network connectivity is very low - RDSI (via DaShNet) utilises AARNET3 at speeds 10Gbps+
6. How will I obtain access?
Deakin will facilitate access via an application process aligned with the NCMAS. An evaluation committee will meet quarterly to review and approve application to Deakin’s NCI discretionary share.
7. I am an existing VPAC user, will I be automatically migrated?
No, but we will facilitate this with you. We would like all existing users to have a Deakin application completed to ensure a fair distribution of resources.
8. How large a capacity has Deakin acquired?
A. At present, 3 million core hours per year. They will be allocated at around 650,000 core hours per quarter.
This total amount is under review and may be subject to expansion in a needs driven basis.
9. What are the Deakin University NCI Resource Allocation Committee terms of reference
The Deakin University NCI Resource Allocation Committee (DUNRAC) terms of reference as located in this PDF (83KB).