Recurring Meeting of Cleveland Clinic - Stanford University
Date: January 17, 2017
Time: 1:00 PM EST
Means: Conference call
- Ahmet Erdemir (Cleveland Clinic)
- Joy Ku (Stanford University)
- Henry Kwong (Stanford University)
Freeform discussion on launch of cloud computing.
Immediate Action Items:
- Cleveland Clinic (Ahmet)
- Continue extensive testing of cloud computing interface.
Stanford University (Joy & Henry)
- Launch cloud computing publicly.
- Before the meeting, the Stanford University team has deployed the sample problem for cloud computing in a staging server. Ahmet tested the interface and provided feedback. This meeting focused on resolving a few issues raised in those tests.
- In his tests, Ahmet identified additional files that did not need to be provided directly as they were already in the compressed raw results folder. Henry removed the links to these files in the user interface.
- In response to Ahmet, the Stanford University team will also change the behavior of running the model without changes. The default will be set to run the model with the default model configuration file. Yet, the contents of the file will be displayed (but cannot be edited) for usability. This update may take a day or two.
- There was an issue with the line breaks of the modified configuration file getting removed. Ahmet suspected that this may have been related to certain web browser. The Stanford University team will try to resolve this before public launch by testing different browsers.
- The participants also discussed potential checks to ensure proper inputs when updating the model configuration file. Ahmet mentioned that he intentionally gave non-numeric inputs which resulted the scripts quit without running finite element analysis. Non-physiological entries will likely result in simulations that may not converge and time out. As long as the system is secure, these issues will not be problematic from a users point.
- Ahmet also recommended providing the raw results in a zip file instead of a tar.gz file. This type of compression may be more familiar for the general user base including Windows and Mac users. The Stanford University team will try to implement this before public launch.
- The ways to provide additional information to the user after submission of the job were also discussed. A message can be provide before the user is directed to viewing page of all simulations, i.e., indicating that the user will be notified by e-mail during the course of simulation. This may need to be done after public launch.
- Ahmet also indicated that a feedback link, at the e-mails and at the view simulations page may be useful to engage with the community. The link may direct to a form to provide structured feedback.
- The Stanford University team indicated their desire to put quotas for simulations. They intend to have a total computing time limit for each user. Ahmet recommended that a time-out limit can also be placed on the simulations, e.g. 5 min for the sample problem. The group agreed on 20 min total computing time quota, possibly providing each user 4 to 10 simulations, depending on convergence rate. Some team members may have unlimited access for extensive testing.
Joy would like to launch the cloud computing feature next week. Ahmet's goal is to get the users' first experiences smooth so that they will return back. Once launched, the feature can be promoted in Biomch-L, LinkedIn and IMAG/MSM Meeting.