Hi Deirdre,
We have a slightly different (more heterogenous) backlog at my institution, but also many TBs of files to ingest and with several processing configurations needed (normalization vs. no normalization, extracting files from archives and disk images or leaving them alone, etc.) depending on the source/type of data.
To accommodate this, we're just now moving to a setup of four Archivematica servers -- one running a central/shared Storage Service, and three processing pipelines. Each of the processing pipelines will be setup with
automation tools scripts so that they're capable of continually ingesting data, with a different processing configuration per server and calling some pre-transfer scripts that do things like grab metadata from our CMS and write it into each AIP's METS file.
This is a step forward for us, but we're already successfully been using several servers in parallel for a year or so now, so I can attest to the fact that Archivematica is capable of scaling up that way. Scaling issues are more likely to happen if you anticipate transfers/SIPs to contain many tens or hundreds of thousands of file each (which is something I believe Artefactual is working on as well). If your concern is size of files, or amount of material to be transferred through in a given amount of time, you can definitely configure your setup to spread throughput across several servers and shouldn't run into issues.
Cheers,
Tim
Tim Walsh
Digital Archivist
Canadian Centre for Architecture