2017-12-15 Tech Call
APTrust Technical Update - December 15, 2017
- Andrew Diamond (Host) - APTrust
- Terry Brady - Georgetown University
- Nathan Tallman - Penn State
- Linda Newman - University of Cincinnati
- Dave Pcolar - DPN
- Christian Dahlhausen - APTrust
- Kelly Cobb - APTrust
- Tim DiLauro - Johns Hopkins University
- Suzanne Chase - Georgetown University
- Salwa Ismail - Georgetown University
- (and 9 more - neither Zoom nor the host captured the entire list)
Slides & Audio
- Request from Nathan Tallman: Restored APTrust bags should include information about files that were once part of the bag, but have been deleted. Currently this information exists only in the PREMIS events in our registry, so you’d have to log in to the Web UI or the REST API to get it.
- Request from Nathan Tallman: Restored APTrust bags should include information about files that changed since the initial ingest of the bag. (Files change when you upload a new version of a bag. These changes generate new “ingestion” PREMIS events.) Currently this information exists only in our registry, so you’d have to log in to the Web UI or the REST API to get it.
- Linda Newman: We should expose the upcoming “single file download” feature through the REST API as well as the Web UI. In general, any feature available through the Web UI should also be available through the REST APIApplication Programming Interface.
- Andrew Diamond: Concurs. Items currently not exposed include Delete and Restore.
- Linda Newman: The Web UI should implement two-factor authentication to make unauthorized access more difficult. The API may at some point need to increase security as well, especially once it permits file deletion and single file restoration.
- Andrew Diamond: The fact that institutional admins now have to approve deletions through an email confirmation mitigates some risk. That does not mitigate the risk of an attacker restoring a single file and then getting access to it. We may need increased security for some API calls to mitigate against that risk. E.g. API keys are valid only from whitelisted IP addresses. Object-level restore is more secure than file-level restore, because an attacker would need access to the S3 restoration to download the restored object. For file-level restore, the system may just give a temporary URL to access the item. (That’s to be determined).
- Salwa Ismail: We need to resolve differences between priorities that the Governing Board has marked as urgent on its list of 2018 technical milestones, and the features that APTrust members have requested on the Issues and Feature Requests spreadsheet.
- Linda Newman: PharosPharos is APTrusts web interface to manage deposits and inspect deposit outcomesETag The entity tag is a hash of the object. The ETag reflects changes only to the contents of an object, not its metadata. (the APTrust Web UI and REST APIApplication Programming Interface) needs a product owner to help define and prioritize what work will be done. Possible candidates: Linda Newman and Greg Raschke.
- Andrew Diamond: Bradley Daigle has been unofficially filling this role for the past three months or so.
- Andrew Diamond: AWS has a new service called Macie that scans data in S3 buckets for sensitive information. This may help us identify what existing material should be encrypted once we implement encryption.
- Several members: Interested in learning more about Macie and seeing what it can do, but it probably cannot identify all types of sensitive information (health info, photo scans, intellectual property subject to export controls).
- Linda Newman: Cincinnati cannot deposit certain types of data until APTrust implements regular security audits.
- Cincinnati and Penn State to send documents describing internal security needs/policies so APTrust knows what to look for in regular audits and what security requirements we must meet.
- Several members said the updates and discussion were helpful. Linda Newman and Salwa Ismail said it may be helpful to hold the meetings every 6-8 weeks.