My homelab setup
A bit of an intro to my homelab
April 16, 2024 — posted by 
My Homelab Setup
Quick Intro
I've been running my own personal homelab for quite a few years now. Looking back, it's been both a pleasant and tretorous experience. I'll begin with my network storage:
My primary homelab storage server contains about 96 terabytes of raw storage, on an ASRock Rack X470D4U. This amount of spinning rust isn't exactly efficient - but I got them at a pretty good price from the folks over @ ServerPartDeals (not affiliated). Not the cleanest job, but it works well enough, and hasn't failed yet*.
I'll explain the rest of my server-related affairs later. My network consists of - as of November 2023:
- Bell GigaHub (10G)
- TP-Link ER8411 Router (10G)
- TP-Link SX-3008F Switch (10G)
- TP-Link SX-1008 Switch (10G)
I am currently waiting for the MinisForum MS-01 to become generally available. Hopefully, it will be able to run 40G and open firmware, as opposed to a locked-down version of OpenWrt (TP-Link).
The 1st iteration
Prior to moving to a server chassis, I was essentially using a PC-case. The first iteration of my NAS contained a bunch of HGST UltraStar drives, which were purchased through a guy (through a guy he knew) on Facebook Marketplace. They were cheap, and rightly so - they were pulled from a working environment, with aproximately 35,000 hours on average out of the eight drives I purchased. I eventually purchased a few more for spares, given the RAID (ZFS) configuration.
Around 2 years ago, the drives cascaded, and... I gave up and purchased new drives. Those drives are still in operation today (model: Toshiba MG07ACA12TE), serving as both my "Linux ISO" storage, as well as a hot backup for my home computers.
Backup situation
You might ask: how did you restore the array once the initial set of drives had failed? The answer is simple: I didn't. I restored a weekly backup from BackBlaze, which is a mirror of my NAS' files. Having said that, I run weekly backups to a VPS, with a mirror to BackBlaze whenever files are updated/removed/etc.
Second iteration
The second iteration, while messy, contains a total of around 98 TB of raw storage.
- Two one-terabyte SATA SSDs for game/quick-access storage (in ZFS' raidz1-0)
- 96 terabytes in a raidz2-0 configuration.
I've attached a photo of it below:
Obviously, it still uses a bunch of consumer hardware (80 Plus Bronze PSU, mismatched SATA cables, Rosewill server chassis). It's not perfect, but it's quintessentially mine (lol).
Compute
The compute situation is more dire. I'm running two 1U servers, configured with Ryzen 5 5600s in each w/ 32 GB of DDR4. Unfortunately, the final (3rd) compute node is a Lenovo Thinkbook, equipped with an AMD Ryzen 7 6800U processor. The limited amount of RAM on the last node makes it difficult to use beyond video transcoding tasks, and doesn't really run virtual machines all that well. In all:
- 2x 1U Ryzen servers
- Processor: Ryzen 5 5600
- RAM: 32 GB DDR4
- Storage: 256 GB NVMe SSD
- Network: 10G
- Lenovo ThinkBook
- Processor: Ryzen 7 6800U
- RAM: 16 GB LPDDR5
- Storage: 512 GB NVMe SSD
- Network: 1G