Post B6irUWpHtNIbagN5sm by wizardmanperson@poa.st
(DIR) More posts by wizardmanperson@poa.st
(DIR) Post #B6inNeF5WN4SkPu3ai by deprecated_ii@poa.st
0 likes, 0 repeats
I've been doing database shit forever but now I'm having to learn about the "big data" side of things and it's mostly pretty retarded
(DIR) Post #B6ipZXhlqv0Br77QTw by wizardmanperson@poa.st
0 likes, 0 repeats
@deprecated_ii in what sense?
(DIR) Post #B6ipZXwexYuabI5Jgm by deprecated_ii@poa.st
0 likes, 0 repeats
@wizardmanperson the pattern goes like "in a normal relational database you would do X, but our dumbass big data way is the exact opposite because it's faster" with no proof or rationale provided that it is, in fact, fasterthis guy just said you should take timestamps and store them in individual token columns like [day, month, year, hour, minute, quarter, day_of_week] because that requires less "calculation" to get the format you want. not only does this obviously bloat up the database significantly, but I'm confident it's much slower to assemble a date string from a bunch of database columns than just getting the timestamp in a universal format, creating a datetime object and calling a string formatting methodbasically these people seem like they have no idea what they're talking about. they just assert things that are counterintuitive and probably false while providing no evidence for their claims
(DIR) Post #B6ipxumzg3zQOsBQu0 by deprecated_ii@poa.st
0 likes, 0 repeats
@wizardmanperson also they don't seem to know about things like materialized views, caching, etcI'm far from a database expert but so far most of the big data "problems" they've talked about are things you can do trivially in postgresql
(DIR) Post #B6iq1s6HcHzhDe5nU0 by Humpleupagus@eveningzoo.club
1 likes, 0 repeats
Why wouldn't you just throw it through e.g. date going in or coming out? If using it as an index, wouldn't storing it in posix seconds be most efficient?
(DIR) Post #B6ir5klcvHhraWqTvU by deprecated_ii@poa.st
1 likes, 0 repeats
@Humpleupagus @wizardmanperson normally databases have a timestamp type which is 4-8 bytes, storing the time in numeric format with time zone datasince the bottleneck is pretty much always going to be reading the database, it makes sense to keep the egress data as small as possible and do any calculations client side. the general rule is if you can calculate it, don't store it
(DIR) Post #B6irU3sTWRcLoAdjKC by wizardmanperson@poa.st
1 likes, 0 repeats
@deprecated_ii amazing. You're seeing this where? Industry "people", online courses (saar), or influencers with big following? I remember the term big data being used a lot some years ago
(DIR) Post #B6irU49UVBEEewbJqa by deprecated_ii@poa.st
0 likes, 0 repeats
@wizardmanperson this is from a linkedin learning course. he's white; I can't even understand the indians without focusing intenselya lot of people in the "data" world have little to no background in programming or computer science, which leads to some really wacky shit
(DIR) Post #B6irUWpHtNIbagN5sm by wizardmanperson@poa.st
1 likes, 0 repeats
@Humpleupagus @deprecated_ii yes, ofc. They're talking out of their ass. Probably retards that want to make a hot viral take to farm engagement, idk
(DIR) Post #B6irrxlzs10vpF0Tyq by professionalbigot69@poa.st
0 likes, 0 repeats
big data niggas hate ISO 8601?that's weird
(DIR) Post #B6jg90dDUWuzK0Dnai by dowitcher@poa.st
1 likes, 0 repeats
this book is highly recommended by non retards https://www.kimballgroup.com/data-warehouse-business-i...