3 ms·
I think we as a profession have made a fundamental error of not internalizing the idea that we have different “colors” of data that need to be treated as such a
by hinkley 5y ago
I think we as a profession have made a fundamental error of not internalizing the idea that we have different “colors” of data that need to be treated as such at all times. At rest, and in motion.
We have bespoke solutions to keep passwords and numbers out of logs by obscuring certain key, value pairs, but that’s exactly what it is. Bespoke.
Those fields should be protected at all levels. I don’t know if I would go so far as calling it a cross cutting concern, but there is definitely a problem with stringly typed data that is a mix of PII, privileged data and common knowledge.
- bob1029 5y agoWe've started to treat anything that even sounds like PII as if it were high level radioactive waste. We have a single unified model for our problem domain with special attributes on those properties which are PII-sensitive. Any time our model is to be exposed to an unsecure context, it is reflected for these PII attributes and mapped into a special redacted variant of the same model. For purposes of troubleshooting, the redacted model properties receive the sensitive data as a hash after it has been passed through salted SHA256. This allows for us to correlate sensitive things like SSNs between multiple log entries for the same work item, but unable to correlate across different work items.
- codingdave 5y agoAbout half the places I've worked, and all the place with more than a couple dozen employees, have had formal security levels on emails, data, and documents. It is common enough practice that plugins exist to set the levels in MS Office tools. These covers PII as well as confidentiality and simply "internal only" levels of content. If you haven't worked in a large company in recent years, maybe you haven't seen it, but it feels fairly standard these days.