OCF has teamed with IBM and DDN to provide a secure, offsite, replicated data archive for the University of Cambridge. The system, which copies data from existing department-level backup systems to provide secondary storage and protection, gives researchers greater confidence in the security and safety of data held. It also enables compliance with data protection guidelines from UK Research Councils funding agencies.
The introduction of a faster network gave us the opportunity to introduce a central storage facility,” said Bob Dowling, head of Unix systems division, University of Cambridge computing service. “The schools funded the facility, but we centrally manage it. Right now, we’ve filled around 10% of the storage capacity and its sized to cope with a five-year lifespan. From an IT management perspective, the system has given us confidence in the security of our data, but also compliance with the grant awarding process of UK Research Councils.”
Managed by the University Information Services, the new storage system extracts data from faculty level backup systems on flexible terms set by the technical staff for each school – this could be twice per day, nightly or on a rotational basis. In some departments the new storage system is second or even third level protection. Right now, 11 departments are actively using the storage facility.
- 400 TB of storage
- 4x DDN 1st generation modular mid-tier storage appliances (x2 at each site) with GPFS file system
- 4x IBM x3650 M4 GPFS (NSD) servers
- Replicated over two sites
The 400TB of storage is contained on a pair of DDN SFA modular mid-tier storage appliances with expansion enclosures. Leveraging 15 years of experience supporting performance workloads and business-critical environments, DDN storage is designed to combine sustainable, low-latency performance with enterprise availability and protection to meet the demands of scalable and latency sensitive applications and workloads.
The DDN arrays are directly attached by FC to a pair of IBM x3650 M4 GPFS Metadata servers, which present the file system out over the Ethernet networks. The solution is also replicated to an identical set of storage on the secondary site. In the event of a primary site failure, the administrator can failover traffic to the replica on the secondary site for availability whilst issues at the primary site are addressed.