Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

For best performance and ease of management, is it better to use 192.168.x.x or 10.x.x.x range of addresses for a small (<50 PCs) business network?

This would correspond to a subnet of and respectively.

share|improve this question
Don't forget, there is nothing stopping you from using – David Aug 11 '09 at 14:18

12 Answers 12

RFC 1918 may offer you some guidance on this. At the end of it all though, you've got to design your network to suit, well, your network. If you've only got ~50 devices on the network, then any /24 ( netmask) will more then suffice

What do you estimate your growth to be? Will a /24 suit in the future? That's a consideration you need to take into account

The performance side of things, I'd probably leave any performance impact to be theoretical. And management, well, that may depend on the tool used. Are you just considering IP allocation. Will DHCP work for you? Do you plan on having statically configured devices? You IP address management may start in the form of a spreadsheet. It may be more elaborate, and be database backed. You may want to tie it in to a NMS or something. There's a lot of scope in this area

share|improve this answer
+1 for actually mentioning RFC1918 – Jeremy Bouse Aug 11 '09 at 13:31

Just because the "default" mask for a space is something large doesn't mean you have to use that mask with that space.

I would pick a subnet in the 10/8 space, like This would give you room to grow in the future (ie add for new space in the current site, and for an alternate site).

For larger initial sites, we usually use a /20 network -- that would give you through to play with, or around 4094 individual IP addresses. That way you can allocate DHCP scopes in logical chunks (like through while allocating other specific addresses in specific places (ie we always put printers and networking gear in 10.0.15).

I would avoid using the whole 10/8 at once because if your network ever grows beyond around 4000 systems the arp noise is going to start taking a non-trivial amount of bandwidth.

I would avoid 192.168.0/24 and 192.168.1/24 because these are defaults for many consumer-grade home devices, and should you ever have to get into VPN access it will cause problems if your users home networks conflict with the "corporate" one.

I personally would avoid 192.168 totally because it doesn't flow off the fingers like the low-hanging fruit in 10.x does. On the other hand, if you have a site policy of avoiding 192.168, it makes it easy to use such addresses for local playing (ie VMware).

share|improve this answer
'it doesn't flow off the fingers' exactly how I see it too. – fduff Mar 10 '15 at 8:33
That argument against 192.x.x.x is very true. I've come across home/work clashes in several cases because a work network used 192.168.1.x – chriscowley May 9 '15 at 7:37
I have just that problem at work. We use the 10.1 subset but some client sites exposed within our intranet use 192.168, the result being that when I work from home I can resolve all domains but cannot access some sites without doing SSH-in-RDP. – Thomas Jul 28 '15 at 20:38

There will be no difference for a such small network.
Just note that 172.16/12 is also reserved for private use. ( to (see RFC1918)

share|improve this answer
+1 I also use for home-installations – ThorstenS Aug 11 '09 at 15:18

As everyone has mentioned, there is no difference between them.

You can carve the address space as small or as large as you want. You want as small as you need, but not so small to make it hard to expand.

The only reason to pick one over the other is if you connect to another network, either by a VPN or by a direct link. You will run into trouble if you have the same address range, so consider what networks you are likely to connect to. Readdressing is not a task to be taken lightly.

share|improve this answer
Sums up my feelings on the subject. +1 – John Gardeniers Aug 11 '09 at 11:18
Pick something like IMO. Chances are the networks you will have to VPN to will be, 172.16 is also a nice touch. – Kyle Hodgson Aug 11 '09 at 11:41

Use 10.X.X.X it is often generally faster to type than 192.168.X.X. Other than that, there is no different except the number of supported hosts, not going into VLANs.

share|improve this answer
+1 for ease of typing ;-) – Dayton Brown Aug 11 '09 at 14:22

There is no difference at all for such a small network - certainly any performance differences would be highly theoretical and miniscule.

share|improve this answer
Not even theoretical. There is no difference. – John Gardeniers Aug 11 '09 at 11:15
I said that because there's bound to be someone who will argue that there will be a TINY difference if maths/processing used for one over the other - people can be like that, but you're right. – Chopper3 Aug 11 '09 at 11:47
this is not true at all, every network tool that will try to discover hosts on the net by scan will try to perform a scan of the entire net. This leeds to geological scan times even for a very small number of effectively present hosts if the netmask used is too "permissive". LEt choose the smallest net you can live with! – drAlberT Aug 11 '09 at 11:48
great point AlberT. Some examples are finding out network printers via broadcast. – hayalci Aug 11 '09 at 13:11

If you have less than 50 PC use a /24 mask. I would go for an series out of convention. 192.168.x.x is reserved for class C subnet and for a /24 that would be appropriate.


share|improve this answer

There's no obvious advantage or disadvantage choosing a range out of, or these days (with almost everything capable of using CIDR). If you're looking at a corporate merger down the line, there MAY be some advantage of picking a random starting point in one of the two larger blocks, as that MAY mean you won't have to renumber. However, planning for that isn't really a priority.

share|improve this answer

It does not mather what network you use no, but there is a advatage in NOT using 10.0.0.X or 192.16.45.x. If you use those networks it can easly create issues if you for some reason need a VPN system, and the network you dail has the same network mask as you.

share|improve this answer

That depend on your need :
192.168.x.x is a Private Internet address Class C that support 65534 hosts
10.x.x.x is a Private Internet address Class A that support 16777214 hosts.

In my network i have about 1000 targets and i use the Class C of address.
May be for some security view you can use 10.x.x.x/24 if you have less that 50 targets. please find here a similar post

share|improve this answer
Almost. 256 (or 254) Class C networks, each able to support 254 devices. However, the actual "class" of network is mostly a moot point in this day and age. – Vatine Aug 11 '09 at 17:44
Indeed, CIDR (Classless Inter-Domain Routing) was promulgated something like 15 years ago. – pjz Feb 10 '12 at 15:11

Either/or, but just make sure that you throw everything you may have ever heard about classes out of the nearest window and use CIDR instead.

With 50 hosts you might even be able to use a 25 bit netmask which would help avoid the VPN issues mentioned by others (I doubt if many networks out there use 25 bit netmasks).

share|improve this answer

The real reason to go with 10.x.x.x for a small business network is, as David touches on, is VPN access to the system.

192.168.1.x is a very commonly used network. Avoid it.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.