sylver_dragon

Have you considered just beige boxing a server yourself? My home server is a mini-ITX board from Asus running a Core i5, 32GB of RAM and a stack of SATA HDDs all stuffed in a smaller case. Nothing fancy, just hardware picked to fulfill my needs.

Limiting yourself to bespoke systems means limiting yourself to what someone else wanted to build. The main downside to building it yourself is ensuring hardware comparability with the OS/software you want to run. If you are willing to take that on, you can tailor your server to just what you want.

I didn’t actually think about what all these wild AV systems could do, but that’s incredibly broad access.

Always has been. I’ve clean Symantec A/V off way too many systems in my time, post BSOD. That crap came pre-loaded on so many systems, and then borked them. The problem is, that in order to actually protect system from malware, the A/V has to have full, kernel level access. So, when it goes sideways, it usually takes the system down. I’ve seen BSODs caused by just about every vendor’s A/V or EDR product. Shit happens. Everyone makes mistakes, but when that mistake is in A/V or EDR, it usually means a BSOD.

Maybe I’m just old, but it always strikes me as odd that you’d spend so much money on that much intrusive power that on a good day slows your machines down and on a bad day this happens.
I get that Users are stupid. But maybe you shouldn’t let users install anything. And maybe your machines shouldn’t have access to things that can give them malware. Some times, you don’t need everything connected to a network.

It’s tough. The Internet and access to networks provides some pretty good advantages to users. But, it also means users making mistakes and executing malware. And much of the malware now is targeted at user level access; so, you can’t even prevent malware by denying local admin/root. Ransomware and infostealers don’t need it. A/V ends up being a bit of a backstop to some of that. Sure, it mostly is a waste of resources and can break stuff when things go bad. But, it can also catch ransomware or alert network defenders to infostealers. And either of those can result in a really, really bad day. A ransomed network is a nightmare. And credentials being stolen and not known about can lead to all kinds of bad stuff. If A/V catches or alerts you to just one or two of those events and lets you take action early, it may pay for itself (even with this sort of FUBAR situation) several times over.

Oddly, one of CrowdStrike’s selling point is that it provides pretty good EDR for Linux and Mac. If you want crap EDR, which pushes you towards Windows, Microsoft Defender for Endpoint is the ticket.

I do agree with what you are saying, but for a complete beginner, and a very general overview, I didn’t want to complicate things too much. I personally run my own stuff in containers and am behind CG-NAT (it’s why I gave it a mention).

That said, if you really wanted to give the new user that advice, go for it. Rather than just nit pick and do the “but actshuly” bit, start adding that info and point out how the person should do it and what to consider. Build, instead of just tearing down.

No, but you are the target of bots scanning for known exploits. The time between an exploit being announced and threat actors adding it to commodity bot kits is incredibly short these days. I work in Incident Response and seeing wp-content in the URL of an attack is nearly a daily occurrence. Sure, for whatever random software you have running on your normal PC, it’s probably less of an issue. Once you open a system up to the internet and constant scanning and attack by commodity malware, falling out of date quickly opens your system to exploit.

Not saying Windows isn’t trash, but considering what CrowdStrike’s software is, they could have bricked Mac or Linux just as hard. The CrowdStrike agent has pretty broad access to modify and block execution of system files. Nuke a few of the wrong files, and any OS is going to grind to a halt.

Short answer: yes, you can self-host on any computer connected to your network.

Longer answer:
You can, but this is probably not the best way to go about things. The first thing to consider is what you are actually hosting. If you are talking about a website, this means that you are running some sort of web server software 24x7 on your main PC. This will be eating up resources (CPU cycles, RAM) which you may want to dedicated to other processes (e.g. gaming). Also, anything you do on that PC may have a negative impact on the server software you are hosting. Reboot and your server software is now offline. Install something new and you might have a conflict bringing your server software down. Lastly, if your website ever gets hacked, then your main PC also just got hacked, and your life may really suck. This is why you often see things like Raspberry Pis being used for self-hosting. It moves the server software on to separate hardware which can be updated/maintained outside a PC which is used for other purposes. And it gives any attacker on that box one more step to cross before owning your main PC. Granted, it’s a small step, but the goal there is to slow them down as much as possible.

That said, the process is generally straight forward. Though, there will be some variations depending on what you are hosting (e.g. webserver, nextcloud, plex, etc.) And, your ISP can throw a massive monkey wrench in the whole thing, if they use CG-NAT. I would also warn you that, once you have a presence on the internet, you will need to consider the security implications to whatever it is you are hosting. With the most important security recommendation being “install your updates”. And not just OS updates, but keeping all software up to date. And, if you host WordPress, you need to stay on top of plugin and theme updates as well. In short, if it’s running on your system, it needs to stay up to date.

The process generally looks something like:

• Install your updates.
• Install the server software.
• Apply updates to the software (the installer may be an outdated version).
• Apply security hardening based on guides from the software vendor.
• Configure your firewall to forward the required ports (and only the required ports) from the WAN side to the server.
• Figure out your external IP address.
• Try accessing the service from the outside.

Optionally, you may want to consider using a Dynamic DNS service (DDNS) (e.g. noip.com) to make reaching your server easier. But, this is technically optional, if you’re willing to just use an IP address and manually update things on the fly.

Good luck, and in case I didn’t mention it, install your updates.

The answer to that will be everyone’s favorite “it depends”. Specifically, it depends on everything you are trying to do. I have a fairly minimal setup, I host a WordPress site for my personal blog and I host a NextCloud instance for syncing my photos/documents/etc. I also have to admit that my backup situation is not good (I don’t have a remote backup). So, my costs are pretty minimal:

• $12/year - Domain
• $10/month - Linode/Akamai containers

The Domain fee is obvious, I pay for my own domain. For the containers, I have 2 containers hosted by the bought up husk of Linode. The first is just a Kali container I use for remote scanning and testing (of my own stuff and for work). So, not a necessary cost, but one I like to have. The other is a Wireguard container connecting back to my home network. This is necessary as my ISP makes use of CG-NAT. The short version of that is, I don’t actually have a public IP address on my home network and so have to work around that limitation. I do this by hosting NGinx on the Wireguard container and routing all traffic over a Wireguard VPN back to my home router. The VPN terminates on the outside interface and then traffic on 443/tcp is NAT’d through the firewall to my “server”. I have an NGinx container listening on 443 and based on host headers traffic goes to either the WordPress or NextCloud container which do their magic respectively. I also have a number of services, running in containers, on that server. But, none of those are hosted on the internet. Things like PiHole and Octoprint.

I don’t track costs for electricity, but that should be minimal for my server. The rest of the network equipment is a wash, as I would be using that anyway for home internet. So overall, I pay $11/month in fixed costs and then any upgrades/changes to my server have a one-time capital cost. For example, I just upgraded the CPU in it as it was struggling under the Enshrouded server I was running for my Wife and I.