Gregory Crane
July 26, 2024
Our colleague at Tufts Technology Services found time during his vacation in Japan to look into the issues with Perseus. It turns out that we have been getting hit with a new set of bots irresponsibly scraping data. These put such a strain on the systems that the weblogs began to fill up the virtual machines that run Perseus — something that is not easy to do. Most users interact with different virtual machines as they move around Perseus. When users found that Perseus was not responding, they were hitting virtual machines that had been filled up and were not functioning properly.
Our colleague took a number of steps to ameliorate the situation.
Tufts Technology Services (in its various names and incarnations) has supported Perseus on the Web since David A. Smith created the first web-based version of Perseus in 1995. If we wanted to provide 24/7 support, we would obviously need a different model. I remain grateful to Tufts for providing the support that it has for almost thirty years.
July 23, 2024
All users of the current Perseus Digital Library (Perseus 4: the Hopper) will have experienced frustrating error messages. Tufts runs Perseus on multiple virtual machines. Tufts moved Perseus to new real machines and that may have contributed to the issue (although it is not clear to us why simply moving to new hardware will have caused problems when the virtual machines have not changed). We have, however, also found that these virtual machines have experienced unusual spikes in traffic and the local server logs have actually filled up the local servers, causing them to freeze. Our collaborator on this at Tufts has been very helpful but is on vacation for the next couple of weeks. The Crowdstrike disaster from last week also was a major drain on the Tufts’ system administrators who are here (2,000 Tufts Windows machines were affected). We will do what we can with those who are available as quickly as possible. We apologize for the disruptions that all of us experience.
The Scaife Viewer does not offer all the services to which Perseus users are accustomed but it does provide basic access to a large body of Greek and Latin texts and translations. The Perseus 4 Greek and Roman collection page also contains links to the Scaife versions.
Our main focus is, and has been for the last two years, creating a new version of Perseus, which we consider to be Perseus 6. Where the Scaife Viewer, built on a new code base, provided us with a much more easily expanded framework for publishing core textual data, a NEH Digital Humanities Grant allowed us to develop Beyond Translation and, in so doing, to learn how to integrate many types of data, including classes of annotation (such as Treebanks and Translation Alignments). We view Scaife and Beyond Translation together as a Perseus 5.0. You can see features implemented in Beyond Translation and read a white paper about the work that has been done.
Support from the NEH Humanities Collections and Reference Resources program has provided support to integrate the scalability of the Scaife Viewer with the flexibility of Beyond Translation. We are fifteen months into this new three year Perseus on the Web: preparing for the next Thirty Years.
We are moving as quickly as we can (and have accelerated our work to the extent possible) to complete a working version of Perseus 6. Our lead collaborator, James Tauber of Signum University, has made great progress on a backend that can manage the data in Scaife and Beyond Translation and that can be rapidly expanded. He is moving to the frontend. We hope to begin replacing components of Perseus 4 in the coming month. We are currently planning to test a prototype version of the Perseus Word Study Tool.
Looking further down the line (and beyond what we can do in our current project), the next step for Perseus would be to create what we are calling a Portable Perseus. This would be a version of Perseus based on the simplest technology base possible. This would not even require a database – all links and all visualizations would be pre-computed. The Canadian Endings Project has proposed restricting implementation to widely supported technologies such as HTML5, CSS and Javascript (without dependence on libraries that may cease to run over time). The price would be flexibility: you would only be able to perform functions that we had anticipated and run services that we could implement in this simpler ecosystem. At the same time, we believe we can represent the vast majority of services from Perseus 4 in such a minimal computing framework. Such a version of Perseus could be downloaded and run locally. It would be much faster and would be structured to run for a very long time without needing to be modified. David Mimno first developed Perseus 4 in 2003 and Bridget Almas completed work on the current version ten years later in 2013. Our hope is that a Portable Perseus could run much longer.