The new HFS on-site disk-based backup service

Introduction

The existing tape-based HFS backup service is, on recommendation from the Backup Service Replacement (BuSR) Project, moving to an on-premise, disk-based backup architecture and with it will bring improved performance and service to the backup of the University's information assets. The disk-based architecture is now in place in two data centres, allowing the asynchronous replication of client backups across sites ensuring both speed and security of those backups while all the time retaining online availability of those backups for immediate recovery. Modern back-up technologies including deduplication and compression will deliver efficiency savings in storage while strong encryption of data in transit and at rest will ensure the continued security of client data. Existing users of the legacy backup service will be invited to migrate to the new service, details of which can be found on the link below:

    The migration process to the new service    

A note about data risk

The service holds copies of all data on servers at two physically separate datacentres. There is no offline copy of backed-up data, and the service is not resilient against simultaneous loss of data integrity at both sites.  Customers with critical datasets requiring a higher level of resilience should consider whether additional backup measures are required.

Benefits of the new service

  • The service is now back to a dual-site fully resilient architecture.
  • Client data is encrypted both in transit and at rest.
  • Disk-based storage should offer faster restores.
  • Deduplication will allow deeper versioning of files beyond the current standard 2 versions.
  • Deduplication will allow faster backups in some use cases
  • Backups are time-limited (8-hours) rather than capacity-limited
  • Deduplication and compression will offer improved capacity limits.

Features

The solution continues to be licensed on capacity so there will continue to be both a fair use presumption as well as upper limits on total backup storage; these are yet to be decided, and experience will shape this.

  • Continues to use IBM Spectrum Protect software
  • Initial versioning has been set at 7 versions held for 90 days – this may be increased once we have more insight into how data storage grows
  • Backup window is no longer constrained by capacity, but instead by time – 8 hours per client.
  • Scheduled backups run overnight as previously but now 7 days a week (previously was 6 days a week).
  • There is no distinction between 'server' and 'desktop' systems - all systems get a daily backup schedule
  • Client encrypted data is not compatible with deduplication and compression technologies: encryption in transit and at rest removes the need for client encryption so that the IBM SP client must be able to access the unencrypted data.
  • A modern and highly functional solution for both server and endpoint back-up
  • Improved long term value for money, compared to the benefits received for the annual cost of the current back-up service
  • An enhanced central back-up service, reducing the need for Divisions, Departments and Colleges to seek their own back-up solutions
  • An improved speed of recovery for business critical applications, in the event of Major Incidents that destroy or corrupt data

Migration

The service will continue to use IBM Spectrum Protect client/server software however migrating from the current tape based backups to the new system is not a simple matter and has to be done within our licenced capacity which restricts how much data can be on both systems at the same time. To achieve this there will be two methods of migration:

  1. Resend of data from the client to the new service. The client will be reconfigured to point to the new service and do a new full backup. Access to the old backups would still be possible if required. After an agreed interval after completing the new backup the data on the old service will be deleted. It will not be possible to keep the old data for anywhere near 90 days though. The initial phase of the migration will exclusively use this method with volunteers who are happy to have their old data deleted fairly quickly.
  2. Copy of data from the old servers to the new. This is a time consuming process as it requires data to first be moved off tape onto temporary disk before being replicated to the new service. However it does avoid having to resend the data and also copies previous versions. This method requires a reasonable amount of free licence capacity so won't start until enough capacity has been freed up by moving clients by resending data (the new service uses licenced capacity more efficiently die to deduplication). Because of resource restrictions it will however have to proceed on a server by server basis and you will be told when you are being moved. The second phase of the migration will predominantly use this method.

Please note that

  • Through the course of the migration process, the service will be limited to existing data stores already secured by the legacy tape-based backup service.
  • Backup is licensed on capacity i.e. the amount of primary storage consumed, so there will continue to be both a fair use presumption as well as upper limits on total backup storage; these are yet to be decided, and experience will shape this, but will exceed the current limits.

Get support


Local IT support provides your first line of on-the-spot help

FIND MY LOCAL IT TEAM

 

Common requests and fault reports can be logged using self-service

   USE IT SELF SERVICE      

   LOG A SUPPORT CALL     

VIEW MY SUPPORT CALLS  

 

The Central IT Service Desk is available 24x7 on +44 1865 6 12345

If you do not have access to your Single Sign-On, you can use this form to contact the Service Desk