Are you me?

This is the exact argument I make about updating.

Without knowing, you’re just trading known risks for unknown risks.

Known risks can be managed.

My opinion of the constant update cycle today: it rationalizes management’s liability offloading while also justifying staffing requirements and therefore budget.

Liability and risk are related, but very different things.

I have 53 nearby wireless networks. Lots of interference.

Channel scan charts just look like a bunch of color mixed together.

I used to modify my access points to use non-standard channels to help.

Still, other than copying media files, it’s fine.

I certainly wouldn’t remove existing cables, that’s effort for little to no gain.

Surprisingly, there’s very little anymore that exposes the bandwidth difference between wireless and wired, and I run a media server at home.

I see the performance difference when copying files, but that’s about it.

I have multiple local devices streaming video (over wifi), while the devices that really need a wired connection get one via a single cable and small switches.

For example there’s a cat5e running to a 5 port gigabit switch behind the TV. So the TV gets a drop and anything else I may need there (Apple TV, dedicated media PC, etc).

Currently have 3 of these switches, with about 10 devices directly wired, but only 3 runs from the primary switch.

NAS sits next to my server, so they’re on the same switch with just patch cables. Pretty much everything else get wifi if it supports it.

Don’t buy any kind of drive from aliexpress.

Shop here: https://diskprices.com/ and even then pay attention to the source.

You have backups or replicants of your data, right?

What’s more important is redundancy, as all drives fail.

What’s your replication/redundancy/backup plan look like? That’s more important than “which drive” (other than SMR/CMR), as even enterprise drives fail.

I’ve had consumer drives running 24/7 for 10 years.

Drives are a lot more robust than most people think, but they still fail at seemingly random times, so having backup is crucial.

Backup, backup, backup.

I have an (old) NAS that frankly I don’t trust to not die. Then again, anything can die, so it’s just one component of my local data replication.

I also have my server which is authoritative for all data, which is then duplicated (on schedules) to the NAS and 2 external drives, so I have 3 local copies.

All of these drives are 5+ years old except the primary data drive which is 2.

I like your point about power monitoring - that would be an interesting secondary method for alerting to a possible breach, if you have a consistent power-use profile.

We wouldn’t be here if we hadn’t already answered the second question affirmatively.

I have an Optiplex that at one point I had 4 2.5" drives in, had to use some duct tape and glue, but it worked fine.

I would suspect he’s getting relayed, and I suspect I was too for my test.

Connecting remotely to your home devices is dependent on your home internet connection’s upload speed, which is usually a fraction of the download speed.

Then add the overhead of the VPN (Tailscale) and how direct of a connection it’s able to make.

Then the connection of the device you’re testing from - it may have some bandwidth limitations.

I just did a quick test - copy a specific file from a local server to my phone - just enabling Tailscale made that copy take twice as long, so it’s definitely adding significant bandwidth constraints (could be an Android limitation).

I run a 2019 Dell OptiPlex SFF desktop as my ESXi box - it idles under 20w with multiple Linux and Windows VM’s (4 are standard, besides the ad-hoc ones for testing stuff).

Hard to beat the idle combined with performance when needed. Pi really doesn’t compare.

What are you trying to do?

In the business world, this would be your business requirements. Once you have those then you can spec the technical requirements.

Without having a target, you’ll just be all over the place.

Start with one thing, get that setup, get management for it in place, backup processes, etc.

Then do the next thing.

Iceberg made a great rec - start with Jellyfin. It’s pretty easy, but touches on all sorts of stuff like storage, backups (which media is worth backing up?), etc. Plus it has a high reward - watching what you want, when you want, from almost any device.

As others have said, unless you’re going to be using those CPUs a lot, you probably don’t need the capability.

I run an old (2019) Dell OptiPlex SFF desktop as my server (I also have an ancient NAS). It runs ESXi just fine, with 2 Linux VM’s and 4 Windows VM’s, with 48GB of RAM.

At idle it consumes just under 20 watts. Peak is 80 watts (limited by the power supply).

Those VM’s all run fine - one is for file services and Jellyfin, one is for dedicated DVD ripping and video conversion.

Even when that VM is converting videos, everything else is responsive. Never get a lag on Jellyfin.

Now imagine how much more performance your server is capable of. Many simultaneous VM’s.

Oh, and it’s a really quiet machine.

I’d sooner have this or multiple mini pc’s than any kind of commercial server hardware. Completely different design approaches - servers are designed for running 24/7 with maximum cooling capability along with max performance. Power and noise aren’t really a consideration.

Sounds like you’re talking about a Mesh VPN.

Syncthing doesn’t need a VPN to function - in fact you’re better off not using a VPN as it’s own rules will see the VPN as a LAN connection and sync data across it when your Syncthing rules exclude using your data connection. Maybe that’s what you saw with Netbird’s data usage. I’d be really surprised if Netbird itself used any significant data.

I have about 20 sync jobs per phone - some are allowed to use cell data (photos), others aren’t. When I enable Tailscale on my phone, Syncthing will try to sync all the jobs because it sees the VPN as a LAN.

I run rooted and use a firewall and block VPN there for Syncthing to prevent this.

Not really.

Resilio has a Selective Sync feature, where it keeps an index at each client, and you select which files to sync in the moment. Works very well, I use it to access (mostly) all my media files (but actually any file on my NAS).

I don’t replace Syncthing with it because it’s very memory intense (keeps the index in ram) and notably harder on battery than Syncthing.

But it works very well - it could replace Syncthing if you wanted.

What are your requirements the make Syncthing not acceptable? That will help us offer other solutions.

Two other tools you can use are FolderSync on Android - it just uses standard network protocols (FTP, SFTP, SAMBA, etc), so it’s much harder on battery than Syncthing.

Resilio Sync is similar to Syncthing but works very differently. I use it along with Syncthing.

Check out Resilio.

I’ve used both for 10 years now, rarely have issues with either, and I sync (with Syncthing) hundreds of gigs between about 8 devices, with about 20 different sync jobs, every day.

You do have to configure ST exactly how you want it, and know what that means.

I’ve been able to move the config 3 times now as I’ve migrated systems - you stop the service, copy the config files (ensuring the new system has the same folder structure), then start it there. Not for the faint of heart.

I do think Resilio is a little more robust, but it’s much harder on memory/battery for mobile - so much so I don’t let it just run on my phone and only use it when I want to sync specific files over (using it’s Selective Sync feature).

Disagree all you want - ECC has no bearing outside of high-resiliency databases.

I say this having nearly 4 decades in enterprise - ECC only matters then.

OP is definitely not doing anything requiring ECC, recommending it is just wasting money.

Why do you want ECC? (Hint: unless you’re running a business database dealing with financials, you don’t need it). I’ve run Windows server on desktop hardware since the 90’s with no issues, and today’s hardware is far better than what we had then.

The reason people settle on NUCs and SFF desktops is power. They virtually sip watts.

I don’t usually recommend specifics for someone but rather ideas and ways to look at your requirements, but given your requirements (20 TB), it would be worth considering a commercial NAS, or at least a NAS enclosure running a NAS OS like UnRAID or TrueNAS.

Expansion is generally not something I’d think about for a NAS (though it can be done today). I expand my NAS once a year (swap out one drive) but I keep 3 local copies - so if it failed I can restore locally rather than from a cloud backup.

So your data lives on a NAS, and you can then either run your services there (they mostly support containers, etc these days), but I’d get a NUC or SFF to host that stuff. It makes for nice separation and gives you some flexibility.

Back to SFF and NUC - my last desktop hardware idled at 100 watts. It was visible on my power bill and used more power than my lights or just about any other single device other than heat or stove.

My SFF server idles at just under 20 watts and peaks at 80 when I’m converting videos. It currently has 8Tb of storage, but I could easily get 20 in there, it would just be expensive.

Oh, and a good NAS can spin down drives to save power when idle, which for most of us is like 90% of the time (I have an ancient NAS as redundancy that does this - it idles around 5w).