A short history of the HFS

The HFS began life in 1995 as an answer to a call for a central more unified method of providing backup and long-term archive for departmental and personal data across the University. It started with a desktop backup service and was followed shortly thereafter by a departmental and college server backup service. A long-term archive service followed, as did a long-term, large-scale file repository service for extremely large datasets. This repository service was known as the Hierarchical File Service which was where the name HFS comes from. Although that service has since been retired the HFS name persists as a brand name for the remaining services.

The initial service was provided using IBM’s ADSTAR Distributed Storage Manager (rebranded as Tivoli Storage Manager in 1999 and as IBM Spectrum Protect in 2015 and to become IBM Storage Protect in 2023/24) running on AIX on two IBM RS/6000 servers, 142GB of direct attached disk storage and an IBM 3494 tape library robot with 4 frames and 4 tape drives. Individual tape media capacity was 10GB and the theoretical capacity of the tape library was 10TB. This was all located in the OUCS machine room in Banbury Road.

In the following years there were of course many upgrades, adding additional storage, extra library frames, upgrading tape drives and replacing the servers.

An upgrade in 2002/3 saw a move from directly attached storage to a switched fibre channel (FC) infrastructure. This created a private Storage Area Network (SAN) out of the tape robot, drives, host servers and SAN-attached disk servers. The tape robot was extended, again and for the last time, to 12 frames, the tape drives upgraded again and the direct-attached disk storage was replaced by two disk servers. 

In 2009 a TS3310 LTO tape library was added to be dedicated to the backup of the University’s new Nexus service (Microsoft Exchange and SharePoint). This was in service for about ten years being withdrawn as the Nexus service was migrated from on-premise to the cloud.

A server hardware refresh in 2010 also saw an architectural change as the AIX servers were now virtualised on the IBM POWER servers bringing various benefits.
Also that year an additional Spectrum Protect server was set up to trial allowing backup/restore over the University VPN service. That was successful leading to the extension to all desktop backup servers in 2014 and the withdrawal of the original trial server two years later.

In 2011 The Hierarchical File Server service was retired as usage had dwindled and the long-term archive service provided a better fit for the typical client-use case.

In 2012 the aging and nearly at capacity 3494 tape library was replaced two new IBM TS3500 tape libraries. One, holding the online primary copy of the data, was located in the new University Shared Data Centre in South Parks Road while the other, used for writing the offsite secondary and tertiary copies, was located in the Banbury Road data centre. Disk storage and servers were split between the two data centres using a stretched SAN for the first time providing a degree of resilience against data centre failure.
Later that also saw the start of the Virtual Environment backup service. This was the first time that primary copy backup data was retained on disk rather than being stored on tape.

In 2015 the HFS team looked at providing a dedicated desktop/laptop backup service providing near-continuous backup. They ran a pilot studies with two products – Druva InSync and CrashPlan. The products were favourably received but it was decided before implementation the whole service should be reviewed. IT Services started a Backup Services Enhancement Project in 2016.

2017 saw the introduction of the HFS Portal which provided the ability to register Spectrum Protect nodes and to check on details of Spectrum Protect accounts. This expanded considerably on the features previously available via OUCS Registration.

The Banbury Road data centre was shutdown in 2019. As the architecture had been designed around two nearby data centres and wasn’t suited to being partly located in the JISC data centre at Slough the service became again based in a single data centre.

The Backup Services Enhancement Project, following a lengthy review of the service and a tender process, selected Code42 (as CrashPlan was known at that point) to provide a cloud based near-continuous backup solution for single user laptops and desktops while an on-premise solution for all other systems would once again be provided using IBM Spectrum Protect. However, the new service would be architected with more modern features, storing all data on disk and doing away with tape, and replicating data between two data centres. The migration to these new services took place during 2021/22.

Get support

If you cannot find the solution you need here then we have other ways to get IT support

Get IT support