I’ve been meaning to do a write up on designing my homelab. In my last job, I had access to hardware and some essential networking bits, but now that I’m a Field SE, i’m in a different situation. I have access to internal tools and nested deployments (otherwise called PODs), as well as some Hands-On-Labs deployments. These are great for doing quick demo’s but for continued education purposes, the consensus among SE’s is that “Nothing beats a homelab”.
There are two big reasons behind that thinking. First, how can you talk to customers about the ease of deployment, if you just rely on an automated processes to build an environment for testing. Second, how can you talk to customers about the “Gotcha’s” in doing something, if you don’t get your hands dirty.
So I sat down and thought about the base requirements and I feel would need to cover:
Full SDDC Deployment
- Software Defined Storage & Networking are increasingly important technologies to test and prove. Especially when it comes to talking to customers. People are more worried about these technologies because its simply different. Building the environment is about putting it personally through its paces. Speaking to how it improves the lives of the Engineer’s and Companies who buy it. If you’re a customer or partner, its about understanding it and making situational comparisons on when it makes sense to use it, because there will be situations when it does make sense.
Data Center Configurations
- This falls more to number of NICs in the server and the methods of High Availability of resources. If you are trying to test or learn, you have to make sure you are practicing towards what you would recommend deploying in a Data Center. I can’t recommend deployments I wouldn’t build myself! I know that most people build using the smaller servers because of cost, but when it comes to testing a technology, you have to push it until it breaks. At the end of the day, thats what matters most, how can it potentially break. So build the lab like you don’t want it to.
Upgradability
- Lifehacker did a great article for July Money Management about concentrating on Upgrading. The reason is that spending a little more for quality and value can help out in the end. In this case, its about spending a little more to make it able to upgrade with other portions of the lab (Like 10G Networking!!). If you take into consideration what you would like your homelab to look like 3-5 years from now, you are doing something at scale what a business does with hardware refreshes. In this case, its your own personal business, “My Education inc.”
Obviously some of these ideas cause the price of the lab to go up, but the idea here is do build a small data center, not just test if the functionality works. Thats what the vendor’s QA/QE teams are for. Do you get paid by them? Didn’t think so. At the core you are testing breakabiltiy, configuration best practices, flexibility. If you just want to test functionality, you can do that with a nested lab. Otherwise, lets get serious about this!
So what am I trying to build?
Compute:
- 1U rack height
- 4 or more Cores
- Dual NIC’s
- 1-2 PCIe slots
- Greater than 32GB RAM Max
- IPMI (Dedicated or Shared NIC)
Storage:
- VSAN with the highest consideration for the Hardware Compatibility List (HCL)
- External storage with iSCSI and NFS available
- External Array with VMware vSphere API for Array Integration (VAAI)
- Expandable storage array, either through network (QNAP vJBOD) or direct SAS connections
- I may want to backup to could using a built in service in the SAN/NAS
Network:
- Dual Top-of-Rack Switches, if “stackable” for data plane, even better (Virtual Chassis).
- If dedicated IPMI is available, potentially a third switch for OOB Management.
- Potential for 10G Networking
- This could be met by purchasing servers that come with 10G networking built into the motherboard like the Xeon D-1500 series servers
- Core Router with VPN, Firewall, DDNS and other services available.
Yes the above list of items in each section is a tall order. If you don’t think or dream big, you run the risk of limiting yourself and your goals. I intend to make a lab that some small businesses would dream of for their production use, at least in the way its configured. This is part one as I begin designing, so stick with me. Comments welcome (Though filtered for spam).
Photo Credit: pixabay.com