Twenty’s Plenty: DataShare v2.1 Upload Upgrade

We have upgraded DataShare (to v2.1) to enable HTML5 resumable upload. This means depositors can now use the user-friendly web deposit interface to upload numerous files at once via drag’n’drop. And to upload files up to 15 GB in size, regardless of network ‘blips’.

In fact we have reason to believe it may be possible to upload a 20 GB file this way: in testing, I gave it 2 hours till the progress bar said 100%, and even though the browser then produced an error message instead of the green tick I was hoping for, I found when I retrieved the submission from the Submissions page that I was able to resume, and the file had been added.

*** So our new advice to depositors is: our current Item size limit and file size limit is 20 GB. Files larger than 15 GB may not upload through your browser. If you have files over 15 GB or data totalling over 20 GB which you’d like to share online, please contact the Data Library team to discuss your options. ***

See screenshots below. Once the files have been selected and the upload commenced, the ‘Status’ column shows the percentage uploaded. A 10 GB file may take in the region of 1 hour to upload in this way. 15 GB files have been uploaded with Chrome, Firefox and Internet Explorer using this interface.

Until now, any file over 1 GB had caused browsers difficulties, meaning many prospective depositors were not able to use the web deposit interface, and instead had to email the curation team, arrange to transfer us their files via DropBox, USB or through the Windows network, and then the curator had to transfer these same files to our server, collate the metadata into an XML file, log into the Linux system and run a batch import script. Often with many hiccups concerning permissions, virus checkers and memory along the way. All very time-consuming.

Soon we will begin working on a download upgrade, to integrate a means for users to download much bigger files from DataShare outside of the limitations of HTTP (perhaps using FTP). The aim is to allow some of the datasets we have in the university which are in the region of 100 GB to be shared online in a way that makes it reasonably quick and easy for users to download them. We have depositors queueing up to use this feature. Watch this space.

Further technical detail about both the HTML5 upload feature and plans for an optimised large download release are available on the slides for the presentation I made at Open Repositories 2016 in Dublin this week: http://www.slideshare.net/paulineward/growing-open-data-making-the-sharing-of-xxlsized-research-data-files-online-a-reality-using-edinburgh-datashare .

NewUploadPage

A simple interface invites the depositor to select files to upload.

 

 

Upload15GB

A 15 GB file uploaded via Firefox on Windows and included in a submitted Item.

 

 

A 20 GB file uploaded and included in an incomplete submission.

A 20 GB file uploaded and included in an incomplete submission.

Pauline Ward, Data Library Assistant, University of Edinburgh

Share

Comments are closed.