Lain's Tech

What do you think a real world equivalent to Lain's tech would be? What real world tech would you add? Obviously a PC (Regarding the Navi) What appears to be a server or something on the left The technology wasn't around yet I don't think, but the system to her left looks almost like a mining rig lmao

yeah definitely a personal computer what else lol

She probably did a cluster with whatever piece of hardware she could get. It would be cool if someone here tried to make a cluster of every hardware piece they own

>>49999 That would be pretty cool. wouldn't it! It's an interesting thought, scrambling for all the hardware you can get and clustering all the compatible stuff

>>50010 This makes me think. I have a bunch of hardware I'm not using. Unfortunately the only switches I have are 100MB. But I'm willing to build a slow system just to experiment. Thinkstation S20 (24 GB) 2x HP compaq elite 6300 (32 GB total) HP Pavilion consumer desktop Various laptops and SFF consumer grade PCs and some mobos with who knows what Anyone know where I would begin with such a thing? I did plan on setting up the thinkstation and bridging with one of the HP Compaqs I'm using now but I would like to take it a step further. I've heard of Beowulf cluster or whatever but don't know much beyond that.

>>50014 Depends on what you want to do. Guix (and Cuirass) can use remote build machines so you could just use it as a glorified cluster build farm. Other CI can do it to varying degrees. The main difficulty is not all of them support a pool of remote builders. sourcehut's does too, I think. There's the various meme orchestrators popular nowadays that kind of replaced older job batching stuff (among other things, they're general purpose tools) used for much the same before. The main issue is that most things that aren't CI or orchestrators are very-much purpose-build/specific and so I can't really provide much help without knowing what you'd like to do. Cluster filesystems are more or less reliable, don't waste your time with ceph. Too finicky and pretty big resource requirements.

>>50014 I'd start with a Proxmox cluster. >>50017 is right about Ceph, you don't have the iron to want to bother with it. I run cuirass, jellyfin, radarr/lidarr, rtorrent/flood, and grafana on a pair of R720XDs with 24 cores/128GB memory each. If you spread out the compute you could easily get all kinds of stuff running. It's going to get messy with so many low-power nodes, but it'll be fun.

Plan9 was basically built for stitching a bunch of different computer systems together into a coherent whole.

>>50014 If you ever wind up getting something stitched together make sure to update us! I would love to keep up with the project

>>50026 I will make an attempt after doing some more research. I don't really have any specialized use case. I'd just like to build a machine for general use. It'd be nice to find a solution that can basically combine multiple machines into one homogeneous platform and spread the workload across the machines.