Deleting backups - Request to delete customer data - Data Protection Act 1998

Question

It was brought up today about the customers right to have their data deleted and went on to an interesting talk about backed up data (we have a rolling 93 day window for backups on AWS s3)

I was wondering if/how anyone out there goes about deleting customer data within backups? It would seem that this data protection act covers backed up data too?

How do you go about this in situations such as mine where we have a 73GB nightly backup file created everyday (expanding to 589GB data and 117GB log files) so in theory if this is fully enforceable and includes updates then we'd need to restore 93 backups it'd take:

Restore backup - 3 hours
Delete customer data - 1 minute - 2 hours (dependant on usage)
Backup 50 minutes

(I appreciate that even although this is a large database to me, working in a small company, this is still small in comparison to enterprises!)

So if we made an application automatically do this, it'd take a minimum of [4 hours per backup] * 93 = 372 hours (15 and a half days!) of processing (on a separate server, so we don't affect our live system)

Luckily we haven't had a request like this yet, but my other concern over this is, if the person writing the script to delete the data accidentally deleted part of someone elses data, we'd now have no backup to fall back on! Surely this would go against your SLA of backups?

I look forward to hearing people's views and any evidence of law on this?

Answer 1

Morgan Lewis is a law firm with offices in the UK and they have published information (as of 2012 which is the most recent authoritative publication I can find on the subject) on the ICO's guidance on deleting personal data under the DPA 1998.

According to their legal assessment of the guidance the ICO recognized the difficulty in deleting electronic data under the act as it can still exist in the organisations systems in one form or another (backup records would seem to apply here) and as such they have adopted what they refer to as a "realistic approach" towards the deletion of electronic data on the basis that it is possible to put the data "beyond use" without actually deleting every last trace of the data. The article states that the key findings of the ICO are...

• Where information has been deleted, but where it still exists in the "electronic ether", such data will not be "live data", and therefore data protection compliance issues will not apply to the data, as long as the data controller does not intend to use or access the data again. The ICO draws an analogy with a bag of shredded paper files-it would be possible to reconstitute the information from the shredded paper, but it would be extremely difficult, and it is unlikely that the organisation would have any intention of doing so.

• It is possible for a data controller to put undeleted data "beyond use" if the data controller is not able, or will not attempt, to use the personal data to inform any decision in respect of any individual or in a manner that affects the individual in any way, does not give any other organisation access to the personal data, puts appropriate security measures in place in relation to the data, and commits to permanent deletion of the information if and when it becomes possible.

Based on the above one option that would not require the extraction, decompression, then recompression and storage of a large number of backup archives would be to add some form of data source separate to the existing backup and restoration systems where an index of people who have opted to have their data deleted are recorded. Then if a restoration of backed up data needs to be done after the restoration has been completed the index can be loaded and the records gone through to see if anyone listed in the index has had their personal information restored through backups, and then have it deleted as needed. Given the fact that you state that you have not had to deal with this to date an easier option (given the low chance of restoring data where someone has opted to be deleted) would be to maintain hardcopy records of deletion requests with the minimal amount of data needed to identify the record needing to be deleted and establish a business policy where one of the steps taken after data restoration is to compare these records to the restored database and see if a record which is meant to have been deleted has been restored.

This would work based on my reading of the Act and my reading of the supporting article by the Morgan Lewis law firm and would be unlikely to cause a major hassle being even a manual process as the number of records that would be requested to be deleted prior to the recommended 6 year auto destruction of data timeline would be low to begin with, and when taken with the rarity of needing to restore complete copies of databases from backup archives would wind up reducing it to an exceedingly small level whereby the process could be achieved in a very small amount of time by a pre-defined user going through and searching for the requested data based on the deletion index list to ensure that if it has been restored it is then deleted manually again. Based on what you have stated this would mean that only data deleted in the last 93 days would likely need to be manually deleted again (even less chance as it is a higher chance of restoring a more recent backup) which would present minimal manpower requirements and ensure compliance with the act.

Answer 2

You shouldn't really receive requests to delete personal data if you are following good practices in safeguarding people's data and not storing data longer than you need to.

There is no law within the UK how long you can store personal data for but it does say you should only ever store it as long as you need to.

Source

Personal data processed for any purpose or purposes shall not be kept for longer than is necessary for that purpose or those purposes.

This is the fifth data protection principle. In practice, it means that you will need to:

• review the length of time you keep personal data;
• consider the purpose or purposes you hold the information for in deciding whether (and for how long) to retain it;
• securely delete information that is no longer needed for this purpose or these purposes;
• and update, archive or securely delete information if it goes out of date.

A good practice would be to allow your customers to select how long that retention is, or allow them to delete the data themselves, you could also have a undelete protocol that holds the data for 30days after deletion at which point unless cancelled purges the data into the abyss.

This way you have no concern how long you store data for since customers are given the option to delete the data themselves.