3 ms·

There are fewer than 100 million shipping containers in use today. A database of all of them, their location, their destination, and what they contain is not ev

by Zigurd 4y ago

There are fewer than 100 million shipping containers in use today. A database of all of them, their location, their destination, and what they contain is not even "big data." this is a mole hill built into a hype mountain with some crypto jargon.

ElfinTrousers 4y ago
So let's see. Back of the envelope math time. Let's round to 100 million containers to make the math simpler. 8 bytes for a synthetic primary key. 8 bytes each also for location and destination, referencing a separate table with information about "places" (e.g. loading docks, port terminals, holds of ships), that should be more than enough distinct IDs to give you as fine-grained location identification as you need. Description, let's say 512 bytes of free text. 8 + 8 + 8 + 512 = 536 bytes, let's say with padding and DB overhead of various kinds we round up to the next power of 2, an even 1kb. 1kb * 100 million is then roughly 100 Gb. Now I am aware of the risk of "why don't they just"-ing, but seriously, nobody else did this math and noticed that the whole database could probably fit on a $15 SD card? No, most likely someone did, and kept their mouth shut, because why blow up a perfectly good grift?
- actionfromafar 4y ago
  While a good bashing, it's not really relevant, is it? It's about collaboration, not absolute size of the data. This comment sounds right: https://news.ycombinator.com/reply?id=33802411&goto=threads%3Fid%3Dsuperice%2333802411 https://news.ycombinator.com/reply?id=33802411&goto=threads%...
- JohnBooty 4y ago
  That would let you describe the current location of each container, right? Presumably we also want to describe a history of locations and events for each container so we're talking a few additional orders of magnitude more storage and complexity. But still, your point 100% stands. We're not talking about something that traditional technologies can't handle. Although to be fair it seems like the blockchain fans are trumpeting the "publicly available ledger" aspect and not the "suitability for Big Data" angle... I'm no blockchain booster, but I don't think anybody is claiming that it outperforms a trad DB.
adolph 4y ago
The containers are only the reference to the contents. The contents are only appreciable across borders as their provenance and affidavit of contents. Those claims are made between counterparties and inspected by government, all of which need to understand the source of the claim and the claim.

by Zigurd 4y ago

ElfinTrousers 4y ago

So let's see. Back of the envelope math time. Let's round to 100 million containers to make the math simpler. 8 bytes for a synthetic primary key. 8 bytes each also for location and destination, referencing a separate table with information about "places" (e.g. loading docks, port terminals, holds of ships), that should be more than enough distinct IDs to give you as fine-grained location identification as you need. Description, let's say 512 bytes of free text. 8 + 8 + 8 + 512 = 536 bytes, let's say with padding and DB overhead of various kinds we round up to the next power of 2, an even 1kb. 1kb * 100 million is then roughly 100 Gb. Now I am aware of the risk of "why don't they just"-ing, but seriously, nobody else did this math and noticed that the whole database could probably fit on a $15 SD card? No, most likely someone did, and kept their mouth shut, because why blow up a perfectly good grift?

actionfromafar 4y ago
While a good bashing, it's not really relevant, is it? It's about collaboration, not absolute size of the data. This comment sounds right: https://news.ycombinator.com/reply?id=33802411&goto=threads%3Fid%3Dsuperice%2333802411 https://news.ycombinator.com/reply?id=33802411&goto=threads%...
JohnBooty 4y ago
That would let you describe the current location of each container, right? Presumably we also want to describe a history of locations and events for each container so we're talking a few additional orders of magnitude more storage and complexity. But still, your point 100% stands. We're not talking about something that traditional technologies can't handle. Although to be fair it seems like the blockchain fans are trumpeting the "publicly available ledger" aspect and not the "suitability for Big Data" angle... I'm no blockchain booster, but I don't think anybody is claiming that it outperforms a trad DB.

adolph 4y ago

The containers are only the reference to the contents. The contents are only appreciable across borders as their provenance and affidavit of contents. Those claims are made between counterparties and inspected by government, all of which need to understand the source of the claim and the claim.