Every new field we explore starts with the same basic question: “Where do I start?” Many of us worked with home computers before moving into the world of professional IT. The lucky ones enjoyed a smooth ramp-up. As the field continues to expand and consumer-grade systems require less knowledge to use, that transition becomes rougher. Furthermore, companies have long viewed technical staff as somewhat interchangeable, so they dislike hiring employees that require foundational training. Once upon a time, many companies would assign new hires to front-line help desk roles and allow them to grow into other roles. Nowadays, most treat IT positions like treadmills; an employee starts, works for a while, and then quits or gets fired. Employers expect applicants to show up with the skills necessary to succeed in the job. That means that if you want to improve your chances of finding a satisfying server administrator job without years of hopping through entry-level positions until you find an employer that will train you, you need to take matters into your own hands. You can take the formal education route, if you have the time and money. Either way, having a home lab will never hurt you. Let’s look at some tips to help you get going.
[citationic]
Can’t I Just Use Online Labs?
Let’s give credit where credit is due: online lab offerings have come a long way. I won’t say a single thing to categorically disparage them. You can learn a lot that way. They work especially well for people brand new to their focal technology.
Home labs still have several points of superiority over online labs:
- The companies that build out online learning environments want you to do things the way that they do them. Practice with that will help you. However, every customer that you work with will do at least one thing differently than the lab prepares you for or will accommodate.
- Online labs have a lot of safety nets and parachutes and padded walls. Although I wish nothing bad on you, you really learn best when you really break something and have to put it back together on your own.
- In most cases, online labs have a predetermined and short shelf life. If you start something and get distracted for a few weeks, you’ll have to start over.
- Having taught many courses as a Microsoft Certified Trainer, I can attest that few people retain much from guided experiences. Most benefit from the introduction and absorb the larger concepts, but almost none remember how to retrace what they performed flawlessly in a predefined workflow.
When I was young (also when I was not so young), my father would watch me tinkering with something and comment, “Son, I don’t think the engineers had that use in mind.” I learn by poking and prodding and pushing buttons and pulling levers as they interest me. Online labs and learning environments have their value and place. For genuinely free exploration, you need something of your own.
Do Not Create a Production-Grade Environment
I know you want the fastest and newest processors, banks and banks of gigantic memory modules, 100 gigabit networking, petabytes of fiber-connected storage, and more. I worry about the motivation and mental health of server administrators that don’t want toys like that. If you can afford those things, then go for it. Just remember that, besides the hardware cost, you also need to satisfy its space, cooling, and power needs. If that still hasn’t dissuaded you, remember that all this takes enough effort that companies willingly pay people to do it.
Truly though, I don’t want to talk you out of building a monster home lab if you can. I want you to understand that you don’t need something like that to learn and succeed. I expect most people new to this field to have a tight budget. You won’t get far for free, but you can make a few dollars go a long way.
Look for Cheap Hardware
Through the magic of hardware advancements and virtualization, you can run a highly functional lab on inexpensive equipment. I use a laptop to design scripts, test procedures, and capture the screenshots that you’ll find in my articles. It runs multiple servers in virtual machines simultaneously. I couldn’t run a company on it (well… I probably could, but I prefer to model acceptable practices), but it certainly performs well enough to show and discover the tools and techniques that I would use in business.
Tips for affordable hardware for a home lab:
- Memory quantity matters most. Do not worry about features like error correction code (ECC) or mirrored or spare modules.
- Try to acquire relatively recent model CPUs, but do not worry about core counts or clock speeds or anything like that. Avoid truly budget CPUs (like Intel Core i3), but almost anything else will work. Even though you will run server operating systems and software in your lab, you do not need to buy chips marketed specifically for servers.
- Storage space does matter, but probably not as much as you might think. ISO images often wind up consuming the most space in my lab, but with a fast Internet connection to download on demand, I could safely clear that up. My aforementioned laptop has a 512GB SSD with plenty of free space.
- For storage technology, you can get by with traditional mechanical drives (look for HDD), but the delays can cause a lot of frustration with even a couple of active operating systems. Prefer solid state disks (SSD). SSD won’t give you the same capacity per dollar as HDD, but you really don’t need terabytes for a lab.
- Do not allow used and refurbished equipment to scare you. I have put refurbished hardware into production environments at the same success rate as new. I like warranties and support, so I don’t put live systems on used hardware, but I have no aversion to it in a lab.
- Think outside the big box stores. You can often walk out of a Goodwill Reboot with a perfectly capable home lab system for a couple of hundred dollars. Many organizations, especially public institutions, sell old hardware at surplus.
- Watch out for scams. You can get too cheap. Unscrupulous people will sell faulty or burned out equipment as new or “gently used”. Some business-class hardware requires now-expired licenses to operate or forgotten passwords to log in with no available hardware reset. I have no bullet-proof guidance to avoid these problems. Personally, if I can’t physically inspect an item before buying it, then I only work through resellers that provide an easy way to reclaim money spent on a falsely advertised item.
- As long as it has the power, your personal use system can double as the virtualization host for your lab.
I want you to understand that you do not need flashy, fancy, expensive equipment to explore the technologies found in corporate datacenters. While the industry has largely forgotten it, the distinction between “server” and “workstation” lies in software, not hardware. Hardware has evolved to better suit the needs of one usage or the other, but the software does not care.
One Host is Enough
I love to cluster things. My first print book was about Hyper-V clusters and I have produced stacks of articles on them. I enjoy exploring new ways to make things more highly available. So, even though it hurts me a little to admit it, you can have just one physical computer in your lab. You can create clusters in guest operating systems. Not every hypervisor directly supports that, but you can usually make something work. Consider it part of the challenge.
Consider Your Physically Installed Operating System Wisely
Keep virtualization front and center in your mind: with virtual machines, you have no commitment to guest operating systems. Physical installations need more thought. The decision belongs solely to you. Do what makes you comfortable. You can seek input from others, but do not worry about their approval.
In the past, I was one of those people that periodically wiped out my main personal system and reinstalled everything. I know that practice continues today. Personally, I would only wipe a system if it caught a piece of malware. A few points:
- All the things that you hear about leftover bits of uninstalled software and “bloatware” installed by hardware manufacturers cause more alarm than real problems. Back when home PCs shipped with primitive operating systems, 160 megabyte hard drives, and 1 megabyte of RAM, we had reason to worry. Even on today’s budget systems, not worrying about orphaned registry entries and folders saves you far more time than you’ll ever recover by cleaning them up or reformatting and rebuilding.
- To satisfy the role of a server administrator, you need to know how to fix things. Lazy, uneducated administrators avoid fixing problems by formatting the system drive and reinstalling.
I take this seriously. My home computer has a Windows image that started out as Windows 7 on an HDD many years ago. It has survived three disk replacements and successive upgrades to Windows 8, Windows 8.1, and Windows 10. Someday, I will move it to hardware that can support Windows 11 and keep it going.
As you digest all that, think about the physical operating system that you will use. I use Windows primarily out of comfort and because no other platform runs all my preferred applications. I have the Professional edition, so I can run lab environments in Client Hyper-V. Windows can run alternative hypervisors, such as Oracle VirtualBox, which would run on Windows Home should I ever choose to downgrade my SKU. I like Client Hyper-V again out of familiarity, but also because it has a wide overlap of usage and features with its datacenter-class sibling Hyper-V found on Windows Server and Azure Stack HCI. Windows 10/11 also provides Windows Subsystem for Linux, which allows me to run multiple Linux distributions without the overhead of full virtual machines.
If you have more comfort with a Linux distribution, use one as your virtualization host. If you want (and can afford) to have a separate machine that you only use for your lab, then you can base your decision on entirely different factors. For instance, you could install a trial installation of Windows Server as the physical operating system. It will expire after 180 days, at which point you could rebuild it to restart the clock (or take the next release of Windows Server). That would give you practice in backing up and restoring virtual machines. You could use an Insider build of Windows or Windows Server.
Whatever you choose, put some time and thought into it. After a time, you will find it a burden to perform maintenance on the physical operating system since all the interesting stuff happens in the lab virtual machines. Make your physical installation something that you know well.
Enrich Your Lab with Routers and Smart Switches
If you have a few extra dollars, look into an additional router and a smart switch. If you don’t have any extra money, you can set up routing in a virtual machine. Some hypervisors, like Hyper-V (on Windows Server, not Client Hyper-V), provide software-defined networking features that can help you practice with concepts that smart switches support in hardware.
As I mentioned in the opening paragraph, home equipment now includes so many wizards and other assistive features that you don’t need to know much of anything about networking to connect everything that you own. I think of that as a net positive, since most home users don’t need to know anything about networking to get through their lives. Systems administrators need a functioning understanding of networking concepts, especially VLANs and TCP/IP addressing and routing. Routers and smart switches can help you gain that.
Managed Switches in the Lab
Standard unmanaged switches offer some RJ-45 jacks and maybe a couple of fiber connections and nothing else. They have a purpose and serve it well. However, most corporate environments employ managed switches. Larger organizations will have dedicated network administrators, but server administrators often assume responsibility for networks in smaller and mid-sized environments. Even with experts to take care of the switches, it helps server administrators to understand their basics.
Managed switches have many features in common, but you mostly want to get experience with VLANs. If you visit a professional organization with more than a handful of users and on-premises servers, you will almost certainly find that they segregate their network traffic with VLANs. The “why” matters, but I’ll let you discover that (or I’ll talk about it in a later article).
Today, you can find managed switches at prices unthinkably low compared to just a few years ago. Look for vendors such as Netgear and TP-Link. Their interfaces, reliability, and capabilities differ wildly from higher quality (and higher priced) equipment from vendors such as Cisco and Juniper, but that matters most to people choosing a profession in network management. For server administrators, the concepts matter more, and all managed switches offer the major features. Also, manufacturers of cheap managed switches target small businesses that can’t afford to spend thousands of dollars on a dozen switch ports. You will encounter these types of devices.
When shopping for a managed switch for your lab, ensure at the minimum that it has VLAN support (802.1q).
Routers in the Lab
Server administrators rarely need to know much about routing beyond the logistics of IP addresses and subnet masks. I have over twenty years of experience with businesses ranging from tiny to gigantic. Even so, I can tell you what “BGP” stands for, but not what it does or how it works. Dozens of other routing concepts fall in the same bucket. I’m sure that there are hundreds more things in routing that I know nothing about. So, I definitely don’t think that a server administrator needs fancy routing equipment in their lab.
However, I do find value in knowing about VLANs. Unless you want to prevent systems in one VLAN from talking to systems in another entirely, then you’ll need a router. Your home Internet router probably won’t allow you to set up VLANs. So, to get the most out of VLANs, you’ll need an internal router. Some managed switches also offer routing functionality, but do not expect it in the cheap ones.
I also believe that every server administrator needs to understand the relationship between routers, IP addresses, and subnet masks. Setting up multiple subnets at home can help you quickly gain that knowledge.
Since I don’t think you need expert-level routing knowledge, you can use just about anything capable of routing. An old Internet router might do the trick for separating two networks. Windows Server or a Linux distribution can route traffic on a physical machine with two network cards. Those operating systems in a virtual machine can do the same if the host has multiple network cards. You don’t even need a second network card if you virtualize your entire lab on the same host. You can even use a Raspberry Pi to function as a bridge and router between your home wireless network and the managed switch discussed in the previous section.
So, Where’s the How To?
If you were looking for a 1., 2., 3. list of things to do, then you might feel like I titled this article misleadingly. I understand why, but I disagree with that assessment. To really make it as a server administrator, you need to practice architecting environments. Start with your own. This article gave you the seeds of thought that you need to work intelligently through the same design work that we use in multi-site systems. The difference is, in your lab, you can’t make a mistake. You can only create learning opportunities.
P.S.: We’ll have plenty of articles to get you through that “how-to” bit.