Re:You may not need Cloudflare Tunnel. Linux is fine
by marginalia_nu on Saturday April 09, 2022 @08:31AM
(#30967478)
My search engine is self-hosted off hardware with those rough storage specs (well, sans ECC).
I do not regret one bit going for this set-up, although my next step toward scaling this up is indeed going to be some sort of hosted solution, the primary reason being the BTUs and decibels involved in running an actual server from my apartment.
Running a search engine or not, anyone with a serious interest in programming should have a server. I really can't recommend it enough. The ability to run large jobs that chew through hundreds of gigabytes of data for weeks if need be, without considering the cost, that dramatically increases the level of ambition you can have in your projects.
* I had an idea for a re-skinned wikipedia. So I processed all 40 gigabytes of wikipedia articles from an OpenZIM file. Took a week. No big deal. Later I used the same corpus to calculate language data models. That's another several days.
* I had an idea a while back. It required me to screengrab 500,000 websites. So I wrote a python script. Takes anywhere between 0.5-120 seconds per site. It's still not finished by any stretch, but I'm up to about 322,000 screenshots now.
I do not regret one bit going for this set-up, although my next step toward scaling this up is indeed going to be some sort of hosted solution, the primary reason being the BTUs and decibels involved in running an actual server from my apartment.
Running a search engine or not, anyone with a serious interest in programming should have a server. I really can't recommend it enough. The ability to run large jobs that chew through hundreds of gigabytes of data for weeks if need be, without considering the cost, that dramatically increases the level of ambition you can have in your projects.
* I had an idea for a re-skinned wikipedia. So I processed all 40 gigabytes of wikipedia articles from an OpenZIM file. Took a week. No big deal. Later I used the same corpus to calculate language data models. That's another several days.
* I had an idea a while back. It required me to screengrab 500,000 websites. So I wrote a python script. Takes anywhere between 0.5-120 seconds per site. It's still not finished by any stretch, but I'm up to about 322,000 screenshots now.
It's just great.