Alcun Atirutan BBS

I read this toot, then the linked blog, then the tootā€™s comments, and it strongly indicates that people need the things I would be working on if I didnā€™t have to work on the opposite things to make a living.

The system that keeps technology moving in the worst direction extends beyond developer and engineer choice, it is capitalism itself, and while I feel strongly that I have solutions to the software and hardware parts of this problem I donā€™t know how to see them through until we find a way to survive while we pursue them.

https://webperf.social/@ryantownsend/112858150684793128

@requiem to be contrarian (and also Iā€™m on vacation so Iā€™m not gonna share any real numbers) large webapps with complex workflows canā€™t be a ā€œsimple perl or php cgi-bin,ā€ particularly when you get into large databases across many clusters, and total codebases older than 10 years. So much time is spent in querying and compute to render.

The author is correct, however, that most people donā€™t have this problem starting out, for personal sites, for startups, etc.

@requiem I know I missed the point, but I think a better way to phrase the original is is: personal projects donā€™t need to be ā€œweb scaleā€ or over-architected. Locally-run programs in a language that youā€™re good at, are fine, & you donā€™t need ā€œindustry best practicesā€.

The things that we have to do to support a centralized service that hits 5rps for tens of thousands of customers during a workday arenā€™t needed for a personal blog. And so on. Build what you need and enjoy, for yourself.

@mathiasx I think this applies beyond personal projects as well. I think the only companies who need to hyper-scale are well, the three or four hyperscalers. šŸ˜

There is the problem of ā€œnow that your data is spread across a thousand systems how do you manage it without hyperscaleā€, but to me thatā€™s a migration problem that begins with scaling it down.

There is also a social/moral aspect to this in that most of these massive datasets exist for bad reasons to begin with, so if smaller scale solutions donā€™t support things like mass surveillance and theft, Iā€™m OK with that šŸ˜‡

@requiem right before Covid hit. I was making a menu bar application that lets you broadcast on the local network that you wanted to go to lunch and help you find people who also wanted to go to lunch. It would be really cool to have that capability again.

@reconbot that would be cool.

To me a non-routeable internet is not an internet at all, and the reason we are stuck with one seems to be a combination of artificial scarcity and lazy approaches to security.

@requiem ip6 is so freaking cool btw, a wildly different approach to the networking Iā€™m used to and very friendly to a routable internet

@reconbot I honestly think the biggest thing holding-back ipv6 is the ergonomics. IPv4 addresses are compatible with the human brain but ipv6, not so much.

@requiem I think this is mostly a... "Google's doing really well. They have runways and towers and airplanes bring cargo. Lets make our own runways, and towers, and maybe we'll get cargo too." Er, MapReduce and Microservices.
replies
1
announces
0
likes
1

@kazriko

Tired: map reduce
Wired: crap reduce

@requiem Map reduce is a good idea... if you're Google and serving trillions of requests across exabytes of data. The vast majority of applications and companies don't need that. If they just made their crap more efficient they could serve it off a single redundant cluster.