What gets kept in Informatics Systems, and where?

Not all of the “Stuff” sloshing around the lab is the same, and distinguishing between them helps tease out the best place to store things. We use a simple Triangle Diagram (originally proposed by John Trigg of PhaseFour which really just tries to point out that stuff is related, but it’s at different levels of abstraction:

It is quite hard to draw definite lines around things, but I think most people can appreciate that a raw data dump from an instrument is somewhat different from a report to management, or that an experimental write up in word is different from some tabular data in a spreadsheet. The differences between the levels come out in:

• The software that’s used to read the file and interpret the content. Some will require very specific software (e.g. from an instrument vendor), but a PDF or text file can be read by many different things.
• Who might be interested in the data. Again, some files are useful to anyone (for example, a report) but some only useful to certain people with specific training.
• How long your company might want to keep the data, and indeed how long you are realistically able to keep the data. Typically the lower you go, the harder it is to keep something, so if you feel it’s business critical you really need to pay attention to the formats used.

This differentiation can really help in ELN System design. Partly it draws your attention to what needs to be stored in the ELN (typically the “Experiment” write up level), and what can be left in systems e.g. a database or a file server, which can be pointed to from the ELN.

Not everything needs to be stored in the ELN, and indeed it would be unrealistic to expect to be able to do so. The important thing is common keys so you can offer the user a link to more information, and the advent of web-based systems has made this level of “integration” so trivial one sometimes feels a bit of fraud describing it as such.

By building on the storage tools you have in place, and focusing an ELN on Experiments, the resulting system is cheap to run, costs little to acquire, and results in little disruption to existing practices.

You can read the final part of this series here.