University of Cambridge Computer Laboratory - Major data centre works 7-22 December – Maintenance details

University Information Services (UIS) and Estates Division are planning to undertake some significant work on their data centre in order to improve resilience and with the aim to "equip the University to host cutting-edge, high-performance computing technologies". The most disruptive part of this work will take place from 7th to 22nd December 2026, during which time electrical supplies within the data centre will be turned off.

There will be some disruption to departmental IT operations for the duration as most of our infrastructure is in the affected data centre. We are working with UIS to try to keep this impact to a minimum by using temporary power and cooling, but at the time of writing (7 October) the details are not finalised and we have no firm statement from UIS of what they will provide. We will update this page as more information becomes available.

Regardless of the anticipated mitigations, some departmental servers are likely to need to remain turned off for the duration, including:

• Research hardware in the affected datacentre: Morello cluster, tfc-app{10,11}, beara, gola, isaac-4

• Archive storage servers: /auto/archive, jerakeen, berilia

• Backups of filer (snapshots will still be created and stored locally)

• SRCF backup NetApps: echo, enid

This is necessary in order to minimise the requirement for temporary power and cooling.

Details of the impact on other departmental services will be announced nearer the time.

UIS have planned mitigations for their own systems and have stated that "all UIS-run, user-facing services will remain available" - though note that we are aware of some exceptions to this, listed below. UIS further state that: "There may be some short periods of downtime for some services before December as UIS service teams prepare for the WCDC works. Service teams will announce these using their usual processes for communicating IT service maintenance."

We understand at present that UIS anticipates partial or complete unavailability of certain UIS-provided services, including:

• Research Computing Services compute clusters (HPC)

• Research Computing Services storage (Research Data Store, Research File Share and/or Research Cold Store - details under discussion)

• Secure Research Compute Platform (SRCP)

• Colocation of departments' physical servers in the affected data centre - where mitigations are not agreed in advance

The severity of the impact on research has been fed back to UIS. UIS invites staff to contact their project team with any queries.

We do not anticipate the works to result in any additional capacity for the Department to host cutting-edge, high-performance computing technologies in future.

This announcement updates our previous one, the text of which has been removed as new developments have improved our understanding of the impact and intended mitigations, and due to sensitivity about us sharing details of the work.