I'm from belgium, starting to get my way around Cisco and IT networking.
For the moment i'm at chapter 6 discussing subnets.
You got the network part of the bits and the host part.
If you want to create 4 subnets you take the fist 2 bits of the host part.
Than you have 62 addresses free for hosts and 4 subnets.
My question is here: what if for example, your HQ has 80 hosts that need an address.
You have the 62 addresses for one sunet, do you need to use the 2 subnets or can you combine one subnet?
I am glad that my answer was helpful. Subnetting is a topic with lots of aspects and can be confusing, especially in the beginning. As you get more experience and a better understanding it does get easier.
Your understanding is correct in terms of traditional subnetting. As subnetting was developed one of the principles was that within a particular network all subnets would be the same size, and as you state the size is driven by the requirements of the largest subnet. This did lead to some inefficiencies, as you note.
To address these inefficiencies a new approach to subnetting was developed. This is frequently referred to as VLSM, which is Variable Length Subnet Masking. Traditional subnetting with all subnets the same size was simple but not so efficient. VLSM changed the basic principle that all subnets need to be the same size and gave us flexibility and increased efficiency in subnetting by allowing subnets within a network to be different sizes. So in your case we can say that one site with 80 requires a /25 subnet (255.255.255.128) but the site which requires only 15 could then have a subnet of /27 (255.255.255.224) and that the remaining address space could then be used somewhere else.
You still can not say the subnet is from 1 to 100 and reserve 101 through 127. If the mask is /25 then all 127 addresses are in the same subnet. But now you can say that the smaller site can use a mask that more closely reflects its needs. Earlier routing protocols enforced the requirements of Fixed Length Subnet Masking while more recent routing protocols like OSPF and EIGRP support VLSM.
Good luck in learning more about subnetting.
While it would seem logical to say that the subnet was 192.168.0.1 to 192.168.0.100 and the remaining addresses in reserve, that is not how IP subnetting works. IP subnetting works on bit boundaries. If you need at least 80 hosts then you would use the first bit of the fourth octet as the subnet bit and then the seven remaining bits are all host bits within that subnet. It does not work to try to use some of those host bits in a subnet somewhere else.