Regarding #1, each DB is from a few MBs, to many GBs total, mostly blobs, size wise. And there can be many such DBs. Regarding #2, only 1 blob per row. I'm aware of existing best practices to keep large blobs in separate tables, and/or at the end of the row. Regarding #3, these blobs are data, either *sensor* inputs, or *computed* outputs, so it's often written only once, and definitely read later, one or more times. Given the hard 2GB limit(ation), I must split large blobs into several rows, no choice. I probably also shouldn't have 2GB blobs, and find the appropriate threshold to split the blobs. 128MB? More? Less? TBD I guess... I know SQLite isn't exactly ideal for very large blobs, but I definitely want to remain "transactional" with other structured data in the same DB (the blob *owners* in other table(s)).