HACKER Q&A
📣 debaserab2

What do you do with your homelab?


For reasons born purely of intellectual curiosity, I'd like to build a homelab.

But, I'm kind of stumped on what to do with it. Like everyone else, my house is pretty well connected to various different cloud platforms for things like streaming and music. It's hard to imagine that a homelab would be a tangible improvement in ways other than things like privacy, but maybe I'm just unaware. I hear of people running K8s clusters on their homelab, which, while being overkill, I could see as a fun hobby project, but I'm just not sure what the actual purpose would be.

Is there anything you can do with a homelab that doesn't have an ergonomically superior cloud alternative? Perhaps the way I have stated that question is a tad too subjective, but I'd definitely love to hear why I'm wrong -- especially if the argument extends beyond the obvious things like not using a proprietary service or privacy concerns.


  👤 detaro Accepted Answer ✓
It's funny how the term has come to mean "overkill home servers" now, and the lab part has gone somewhat missing or at least isn't a core part anymore. A lab is traditionally where you experiment and try stuff you want to understand, not something you use to run your home network. E.g. for network folks going for vendor certifications often a big pile of outdated (=cheap) networking hardware to get hands-on experience. Or if you want to learn Kubernetes or try a specific configuration and set that up on some servers.

I don't really have one, experimental/learning stuff is done either on my normal PCs or on VPSes. I have a small home server, which just runs filesharing and some tiny services.


👤 anderspitman
You might be surprised how refreshing it can be to host your own alternatives to cloud services. It's a matter of incentives. Cloud services will always evolve to maximize profits. While this is closely tied to customer satisfaction, you might not be the customer. Often advertisers are the true customers.

Open source tools tend to be much more user-centric. Even just dropping dark patterns and constant upselling often makes it worth it for me.

It's not a panacea. OSS has its own issues. But it feels better for my mental health to not be using software that's constantly trying to manipulate me in some way.


👤 c7DJTLrn
I'm skeptical of homelabs. There's people buying old rack mount servers on Ebay thinking they're getting a great deal when really they're getting a 10 year old machine that will obliterate their ears and provide more heating than processing power.

Anything you want to experiment with can be virtualised or ran in the cloud. The only legitimate use case for a homelab I can see is to play with hardware but it sure is an expensive hobby just so you can play around with RJ45 cables and switches.

Even for a NAS, you're just making yourself a system administrator of something you really don't want to go wrong. Save yourself the hassle, put it all in an S3 bucket and spend your hours on something else.


👤 grepfru_it
I have a real lab that is 12 beefy servers in various data centers. My home lab is a smaller version of that which is only 5 not-so-beefy servers. I test all of the things I want to try there before making them somewhat permanent in my data center.

I currently have 5 openstack environments because I wanted to test rapid iteration of those environments. They are all publicly accessible thanks to a /25 I have on my home network. And since it’s my home network I have to maintain some sort of a production stance otherwise the end users will get uppity. So just like in the datacenter, I have a “production” network that is highly available and runs all of the 24/7 services for my house (home automation, file server, etc)

I run all the services you would expect a cloud to offer.. vm workloads, container workloads, serverless functions, orchestration, managed services for things like Java servlets, MySQL or Postgres, s3 layer, web based office apps, Kanban, project mgmt, etc.

This all started as a hobby project to learn software features or experiment enterprise ideas outside of work. I flipped it into a side-gig a decade ago. Adding the side-gig experience to my resume 3x’d my salary over the past decade. The income from services technically made that 4x. I no longer host services and the servers sit 90% idle. They are technically paid for for the next decade but I keep it around in case others want a starting point..


👤 0x0000000
My homelab started out of CCIE studies, so it was originally focused on networking and virtualization. I still use it for this, e.g. running a Cisco ASAv or VeloCloud vEdge on ESXi, I've also run Xen just because I wanted to try it out. This is where I would spin up a sandbox to learn how to use something like Wireguard, or Splunk, or whatever else $dayjob or $interests prompt me to explore.

But I also use my homelab to run "production" services as well: Emby/Plex/Jellyfin, Vaultwarden, a local Wikipedia, and more to come.

The overkill gear let's me do handy things, like split off my IOT devices to their own SSID/VLAN with appropriate firewall rules. I can run a guest wifi network, with rolling password auth, or even no auth at all - in which case I might want to block certain web traffic and such, which I can do with my squid proxy.

In summary, the answer is "whatever I want", and what I want is often to play with new (to me) tech, or practice professional skills.


👤 MrWiffles
The company I work for at the moment does a lot of work with k8s for government and enterprise, and several of my coworkers have home labs set up purely so they can test things across multiple specific versions of k8s, like upgrades, reproducing unusual bugs to figure out a solution for clients, etc.

Personally I don’t have the hardware to do that myself, and I’m thinking I’d making a change in jobs anyway, but I have been building up personal storage and backup devices for storing/serving media. Got two RAID0 (YOLO!) NAS devices, one as the primary and another as a backup target. Not quite a lab, but after losing everything I owned a few years ago, it takes a while to rebuild.

Another co worker has his home lab set up for VMware administration and testing across various versions and hardware configurations. It’s been a life saver for us too, since his knowledge was invaluable on a project back in October.


👤 0xBDB
When I was a systems and network engineer I used a VMWare and GNS3 lab a lot for labbing out network scenarios and testing configs.

These days as a Red Teamer I use VMWare for setting up vulnerable virtual machines and entire Active Directory environments to lab out attack scenarios. Anything that needs to be on the public Internet, like a test vulnerability scanner, probably goes on AWS instead.


👤 k4ch0w
It's used for getting better with new tech. I use them to run crawlers, Jellyfin, Rust builds :-p, code on, experiment, perf test I guess just random things that come up. It's good to understand building and configuring one yourself and how it's made and I also like to have a local backup of my photos.

👤 ddingus
Well, my home lab consists of:

An Apple //e computer with a variety of cool cards in it. Gives me access to 65C02, 65C802 (address limited 65816, and the hardware is 65816), Z80, and at some point 6809, but it looks like I might have to build my own card for that one. I do have a PC type card for 8086/88, but it's convoluted and I may just get an actual PC, maybe one of those industrial ones LGR and others are showing off. Much nicer, more capable.

I have a Model 100 for the 8085.

Couple of other retro systems in storage.

These are all assists to learn tech new to me, reverse engineer, make replacements, etc... The Apple, strangely, is a great 8 bit workstation! I would not have called that so long ago. If a CPU is unclear, I can write a little test harness and then clear things up.

Electronics:

Tek 300Mhz analog, 4 channel scope. Looking hard at a digital one. Various meters, signal generators, tools for electronics work, testing sensors, prototyping systems for use in products.

High end 3 extruder 3D printer. Can do 3 serious polymers, or a mix of support and polymers, flexibles, and the only materials out of range for it are the uber expensive high temp stuff. PEKK and friends, though Victrex did recently release AM 200 which works GREAT on my machine. This is professional work and it's all about making sure others can print parts and meet requirements. I will print stuff, ship it out, and when they buy a machine, I can then pass the education along, get them moving. This is a lot of fun!!

HDMI monitor, 15Khz CRT PVM for classic signals, retro grade, flat CRT VGA will do 2560x1xxx @ 50 / 60Hz. I rarely push it, preferring 1024x768, perhaps 1280x1024. All, except the HDMI display are 4:3.

Microcontrollers: Parallax Propeller, 1 and 2. If I need to build something, these are my GOTO. I know the old CPU's very well, but so far have not needed to build anything. I can just put code on the Apple, and learn what I need to. Ben Eater is tempting me though!! Great work. He is making a big contribution to the field. Lots of people are gonna come up through the ranks like us 70's and 80's kids did. Wonderful! We need those people.

Dev machines running Linux, Mac OS, Windows 10. I added the Mac because I want to follow all the fun on the M1!! (seriously cool chip) Haven't done much with it yet, but I'm going to help port an app to the platform.

Various hand tools, drills, saws, sockets, everything from taking apart some electronics that find it's way in here to working on my car.

I have had some form of this since I was a kid in the 80's literally!!

Back then I had a Tek 80Mhz scope, 2 channel, Apple and Atari machines for development and control projects, various meters, signal generators.

I use the Parallax chip as a poor man's logic analyzer, and it's good to like 20Mhz, and much higher on the newer one.

Stuff comes and goes.

In the 80's, it was about assembly, serial comms, driving CNC gear, controlling things.

In the 90's, it was about networking and I had stocked a bunch of SGI, Linux and WinNT gear. Was running everything on a DSL, Cisco modem, awesome! Learned a lot, then ditched all of that because my professional work went from bigger systems to smaller ones. When it was time to step away, or commit to the data center big IT path, I literally gave most of it away to someone wanting to go down that road.

In the 00's, more programming, product development, building things for research, helping others get things done. For a while, I was working with big enterprise stuff and had a whole environment or two virtualized at home. This was extremely valuable, until I stepped away from that space. Frankly, wanted to go small, where the bits are, one can see the pixels, hear the sounds, sense the world, control things.

That dove tailed right into additive where I am now and am working on some advance projects. Some of it leading edge, likely startup type stuff. Having a home lab, and making the friends I have is the only way I would have ended up here, and being as effective as I currently am. One side effect of this is I can probably fix almost anything, and I can definitely bring gear up and have it performing, validate it, and prep it for whatever task is at hand. This has been very high value.

Most recently, it's about supporting pro work with 3D printers. Got some spiffy machines coming into the country and I'll make sure they can print all the sexy polymers, print sample parts and in general validate any use case people want validated.

And always some programming, building my own tools to use on projects.

Most of this stuff has been paid for by the work I've done in the lab at home.

I strongly recommend this path because it makes one think about the money they spend and how relevant the gear might be. Always find ways.

Get old crap out of dumpster and fix it.

Trade work for gear. (my favorite, makes everyone happy)

Make gear, signal generator, logic analyzer, various test harnesses. If I get good, and realize I need more, then I'll buy, or scrounge and get the proper stuff. Take the harder projects, learn the harder things.

I'm writing this because there is a whole world of development, product design, CAD, assembly, test, measure, that does not have much to do with the cloud.

Above all, put what you love in the lab and do shit. Push it. Find out the real risks, real limits, etc... This generally pays off and I know very few people with bad experiences and they worked to other's expectations and goals, not their own mastery for their own reasons.

Have fun!! If it is not fun, ditch it, get new stuff, until it is. Seriously.


👤 hulitu
I have a wife, i don't have a homelab.