Homelab Updates - Winter 2021

Homelab Updates - Winter 2021

Since my last homelab post, I’ve added quite a few components back into the lab mix.

Lab Hardware

The diagram’s a lot, so here’s a high-level hardware breakdown:

---
switches:
  - name: "Core Lab Switch"
    part_id: "Ubiquiti EdgeSwitch 24 Lite"
    ports:
      1gbps: 24
      1gbps_sfp: 2
      10gbps_sfp: 0
  - name: "Lab PoE Switch"
    part_id: "Ubiquiti EdgeSwitch 8 150W"
    ports:
      1gbps: 8
      1gbps_sfp: 2
  - name: "SAN Network Switch"
    part_id: "Mikrotik CRS305-1G-4S+IN"
    ports:
      1gbps_copper: 1
      1gbps: 4
---
hypervisors:
  - name: "Proxmox"
    proc: "Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz"
    memory: "64GB"
  - name: "Proxmox2"
    proc: "AMD Ryzen 5 3400G with Radeon Vega Graphics"
    memory: "32GB"
  - name: "Proxmox3"
    proc: "Intel(R) Celeron(R) J4105 CPU @ 1.50GHz"
    memory: "16GB"
---
arm_clusterboards:
  - name: "Turing Pi v1 - 1"
    nodes: 7
  - name: "Turing Pi v1 - 2"
    nodes: 3
---
storage:
  - model: "QNAP TS-332X"
    disks:
      - type: "Seagate 4TB"
        count: 3
      - type: "Crucial 500GB M.2"
        count: 3

There’s a lot of new hardware here compared to the one Hypervisor with local storage I was running during summer 2020.

What am I doing with it all

With the standard virtualization, I’ve been primarily using it to run various Kubernetes clusters. Between clusters launched with kubeadm, rke, and k3s, it’s been an excellent learning lab environment. Additionally, I’m hosting virtual machines for a ZeroTier router, Unifi Controller, and PiHole.

The QNAP TS-332X has proven itself as a surprisingly great little NAS, and the SFP+ connectivity with two of the big hypervisors has been fantastic. I can max out network speeds at nearly 4Gbps, or just about 500MBps read/write on the SSDs, or about 2Gbps read/write on the spinning drives at nearly 250MBps. Initially, I was incredibly skeptical of how its performance might look with a smaller ARM processor, but I’ve been quite impressed with the Anapura Labs Alpine AL324 processor in it.

With the Turing Pi cluster boards, I’m actually hosting this website on k3s. It’s been an absolutely awesome little board. I was happy enough with it to buy a second on when they announced the second batch, and I’m actively working on ways to use it with k3s in more edge-compute scenarios. Honestly, it’s one of the more impressive boards I’ve seen in a long time, and I’m excited to pick up the next version when it’s available.

What else is different?

  • I’ve physically re-arranged everything in my office to live on a racking system rather than in a proper server rack.
  • I added a second UPS and a CyberPower ATS PDU. Absolutely fantastic decision.
  • Moving from a single node to a cluster of nodes for virtualization made me remember how nice is is to be able to keep uptime.
  • Most of my recent Kubernetes deployments have been with rke rather than kubeadm, and it’s dramatically changed how quickly major cluster changes can be made.

What’s next?

I’m hoping to have some time in the near future to put together some articles on some of the newer platforms I’ve been playing with; Rancher rke, Terraform, and Longhorn are on the short list.