My consumer-grade router currently uses its defaults for DHCP assignment (192.168.1.0/24)

However, I’d like a “tidier” network. I’m thinking about using 10.0.0.0/21. I’d like to assign IPs for each known user’s device into a respective subnet.

This is entirely unnecessary; I don’t think I will net any tangible benefits. However, doing so would allow me to achieve a few things:

  • I learn about networking. I’m currently a total noob.
  • I am able to build a marginally more organized network (subjective)
    • example: client configuration in AdGuard Home is slightly more sensical to my human brain
IP range Usage
10.0.0.x “Common” devices; router (.1), home server (.2), Roku (.3), printer (.4)
10.0.1.x User A’s devices, manually assigned via MAC address in the router console
10.0.2.x User B’s devices, manually assigned via MAC address in the router console
10.0.3.x User C’s devices, manually assigned via MAC address in the router console
10.0.4.x Automatic DHCP assignment starts here; I would manually assign new and recognized devices (i.e. not guests) an IP in their proper subnet when I get around to it.

Questions

  • Does this make sense?
  • Are there any consequences I am not anticipating?
  • Are there any performance considerations?

Potentially very stupid questions

  • I’ll never actually use all 2,048 addresses under the /21 mask – is there any kind of performance impact by making so many unused host addresses available?
  • I’ll never have 254 devices on this network, let alone 254 on a single subnet. Should I be… “spreading out” the assigned host addresses? Like instead of .1, .2, .3, assign them .8, .16, .32, etc.?

Network information:

  • Verizon FiOS 300Mbps
  • CR1000B router/switch/access point all-in-one
  • Wyse 5070 running a few Docker containers, including a DNS blocker
  • roughly 12 connected devices, no IoT
    • four laptops
    • three smartphones
    • some secondary devices (smart watches)
    • one Roku, used between eight and twelve hours per week
    • one printer, almost always off

I will not be replacing the CR1000B anytime soon. Proper VLANs seem to be out of the question. This router does offer -Guest and -IoT; I think -Guest is VLAN’d with a maximum of ten connected devices (weird arbitrary limit) and the -IoT one is 2.4 GHz only, but not VLAN’d.


I’m a networking noob, so please correct me politely.

  • e60deluxe@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    7 months ago

    Does this make sense?

    not really.

    Are there any consequences I am not anticipating? Are there any performance considerations?

    your IP scanners are gonna run a lot slower…other than that no. large networks are usually a symptom of a larger issue, but not a problem in and of itself. If for example you have 50 devices, putting them on a /24 or /20 or /21 wont likely make a difference. but if you have 1000 devices, deciding to solve that problem by creating a /20 does NOT solve the problem of 1000 devices on the same broadcast domain. but dont conflate those problems with “dont use large network sizes such as /20 or /21”. does that make sense?

    I’ll never have 254 devices on this network, let alone 254 on a single subnet. Should I be… “spreading out” the assigned host addresses? Like instead of .1, .2, .3, assign them .8, .16, .32, etc.?

    Most people do what you are doing but dont increase the network size just to do so.

    for example, instead of 10.0.0.x make it 10.0.0.200-254

    instead of user A’s devices being 10.0.1.x make it 10.0.0.10-19, 20-29 for the next user, 30-39 etc.

    then the DHCP range make 100-199.

    that way you still have equal “tidiness” without needing a humongous network size. but its up to you.

    • Superiorem@alien.topOPB
      link
      fedilink
      English
      arrow-up
      1
      ·
      7 months ago

      Thank you for your thoughts!

      instead of user A’s devices being 10.0.1.x make it 10.0.0.10-19, 20-29 for the next user, 30-39 etc. . . . then the DHCP range make 100-199 . . . that way you still have equal “tidiness” without needing a humongous network size.

      This is what I started with before realizing I could also assign the third octet. I’m glad to hear that my original plan would have worked just fine (if slightly uglier to my eye… but I’ll choose function over aesthetics).


      Some clarifications about what you quoted… I originally wrote:

      Should I be… “spreading out” the assigned host addresses? Like instead of .1, .2, .3, assign them .8, .16, .32, etc.?

      Did you intend to highlight this portion? And if so, could you please explain further?

  • DwarfLegion@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    7 months ago

    Seems like you’ve got a decent plan and have researched enough to make it happen. No major red flags, but here are some notes:

    10.0.0.0 /21 gives you 10.0.0.1 through 10.0.7.254 as usable addresses. You’ve got 8 useable subnets here 10.0.0.0, 10.0.1.0, 10.0.2.0 etc.).

    Therefore, consider 10.0.0.0 /22 which gives you 10.0.0.1 through 10.0.3.254, for a total of 4 subnets, unless that is not enough for your intent.

    Now, understand that having so many extra IPs isn’t going to inherently cause you any issues (unless maybe if you are doing a lot of internal network scanning), but it also isn’t necessary to achieve the logical separation of devices you’re looking for.

    Instead of having different subnets for each device grouping, you can put them all in the same subnet using predefined ranges. For example, X.X.X.1 through X X.X.9 might be servers. .10 through .19 can be dedicated to printers. And so on.

    If you still want the different subnets instead for your own experience, it will also function just fine. It’s just adding a lot of extraneous space.

    As for spreading out host addresses, no need, but you can if it fits your ideal network scheme.