You Don’t Need a Blockchain, You Need a Time-Series Database

I've definitely been around too many business people who want blockchains for no real particular reasons then to get onboard the "innovation".

Most of them just want a ledger. Blockchain is just used now as the "solution" to anyone that wants a ledger.

The issue is, in the financial world, it seems distributed ledger refers specifically to blockchains. So now there is no way to talk about a distributed ledger the way you'd describe a distributed system, that is, a database used as a ledger that is replicated accross site, but which is still owned by a single entity and administrator. And now people say they want a private blockchain, which is basically them saying they just want a normal ledger but they want the database to meet modern database standards of having multi-site replication.

The part about scalability is very outdated. And comparing it to a centralized database is useless anyway. You would need to compare it to global net of nodes which all sync with one master database. Due to the time it need to sync you simply cant ever reach performance of a single DB.

The second thing that is misleading is the idea that a read-only DB would be possible without either trust or decentralization.

Just because your software doesn't implement a delete or update method doesn't mean its read-only at all. And even if there is crypto verification in place like a chain of hashes that would break if something is changed in the past. I all means nothing if you dont wanna trust the source of the data and verification.

The key part of "blockchain" (pls call it distributed ledger (DL), it doesn't need chains of blocks) is the public state of the system. It allows to verify the data without trusting the parties that gave you the data.

The conclusion of the article is correct. If you dont need the trustlessness and decentralization you almost certainly dont need a DL.

Hijacking somewhat: does anyone know of a small time series db library for C? Looking at doing data logging with a RPi and wondering if there's anything better than CSV in terms of file size efficiency

Title from a parallel universe where it wasn't sponsored:

"You don't need a blockchain, you just need to pick one of Postgres, MySQL or SQLite"

Ok this questdb looks cool. Only downside currently is my biggest usecase for one would be gis queries. any plans on supporting gis?

or you need a https://en.wikipedia.org/wiki/Two-phase_locking

Long time an article like this hasn't appeared but it was due because of the PayPal announcement. Anyway, trust is the issue.