= GIMS = The ''GENI Instrumentation and Measurement Service'' overlaps between most (all?) of the WGs. This is a common page to collect the ideas of all concerned. Work on the parts of GIMS relevant to each WG will happen in that WG, but this common page exists to allow anyone working on GIMS to find and/or record concerns that need to be addressed. == List of concerns == These are the collection of concerns with GIMS. Please add any more you think of. * There are many privacy concerns with the measurements. * Some measurements will be "personal information" about users, which has moral implications (and possibly legal ones in some jurisdictions). This impliles that Opt-In users need to agree to this collection. * The level of privacy for given measurements may change over time. For example, when an experiment is first run (and especially while debugging) the researcher will want measurements private, but may want to make them available after the experiment's results are published to allow others to verify the results. * Do measurements require certification, i.e. a signature to verify they haven't been tampered with? Does a trusted repository address this? * Operations will need to collect measurements as well. Can we keep from double collecting and still keep the security and privacy concerns? * How long do measurements need to be archived? One comment (with an illustrative example) from GEC!#1 was data over 5 years back was needed. One observation was "However much storage you can afford." Does the long term archive all need to be online storage? How do we weed out useless data that doesn't need to be kept? Who could make that assessment? * How much does the archive need to record about how measurments were made? Someone suggested in discussion at GEC!#2 that knowing how a slice was set up for a set of measurements might be important. But, that requires that the researchers export that information as normally nobody but the researcher knows which version of the code was loaded. * How do you avoid Heisenberg interference in measurment collection? There have been some suggestions: * Create a separate slice (or just another set of slivers in the same slice) for the measurement plane. * Some components may be able to store measurements and allow transfer after the experiment is over. * Measurements may be transferred to the repository over the legacy Internet.