I would like to ask for example 192.168.1.0/24 which has new ip’s added that are using /30, can existing or new ip’s still use /24 on that subnet?
What are you trying to accomplish? Give us the use case.
That’s not the same subnet though, a subnet is made up of a network and a net ask, if you have 2 separate net asks you have 2 separate subnets.
Originally our WISP network was using OSPF but because of adjacency disconnections on wireless interfaces, we had to start using bridged VLAN’s for PPPoE ,
OSPF is now used for management and just wondering if the the original IP’s + OSPF network setting were correct ?
Example if a one side of PTP was using IP 192.168.1.13/30 network 192.168.1.12/30 and the other side of the PTP was using IP 192.168.1.14/30 network 192.168.1.12/30
Then we have another devices using network /24 - example IP 192.168.1.100/24 network 192.168.1.0/24
In other words can you have IP combinations of /30 + /29 + /28 + /27 + /26 + /25 along with /24 on a network?
is this a bad configuration setup.
It’s not real clear what your saying, but a general rule of thumb is you can not reuse subnet’s, in the same Network, there are some exemptions to that, but not for your case.
EDIT: Just to further break this down the 192.168.1.0/24 subnet contains or “Summarizes” all the other one’s you listed so they are essentially duplicates if you use a /24 with those /30’s etc.
Yep, overlapping subnets are a very bad idea. Not every bit of networking equipment might choke on them, but in terms of network design it is a no-no, especially when there are so many different /24 subnets available both the class A and class B private spaces.
There is no good reason to have overlapping subnets.
I wrote an article on bridged to routed migration for WISPs that may be helpful.
https://stubarea51.net/2019/09/15/wisp-design-migrating-from-bridged-to-routed/
Also this article on overall design might be helpful - VPLS/MPLS is a very popular choice when you need to aggregate PPPoE sessions from towers.
https://stubarea51.net/2020/03/03/starting-a-wisp-guide-to-selecting-a-routing-architecture/
Yes and no. When devices communicate between each other (unicast), they only use IP addresses … or rather MAC addresses. However things likely go haywire when communication peers use different subnet masks and when one device considers peer as member of same subnet while the other one doesn’t (the first device uses shorter netmask, e.g. /25 v.s. /28). I this case device with longer netmask will utilize its gateway while device with shorter will reply directly. But that will only work if the router used knows how to deliver packet to device with shorter netmask. So effectively there’s a triangle … And then there’s issue of broadcasts (some services rely on it; broadcast addresses in subnets with longer netmask are regular IP addresses in subnets with shorter netmasks) and other minor things.
Sp, as many people said: it’s bad idea to use overlapping networks if one doesn’t know exactly why that’s needed and how to overcome all the minor glitches bound to happen in such scenario.
Thanks for replies so far and apologies if I haven’t explained better my question!
As regards migration to a routed network, we had no option a few years ago to migrate from routed to bridged network using vlan’s,
We use a 100% Mikrotik network and OSPF used on wireless interfaces is simply not resilient to interference or signal levels variations, we had too many
OSPF adjacency disconnections with up to 5mins or maybe more depending how many OSPF networks lost adjacency!
We have a more stable network now but YES configuration is very laborious and chances of creating a network loop is high?
Here is a screenshot and my question is 10.0.0.0/8 necessary when most of the /30 in (10.0.0.0 - 10.255.255.255 ) range are disabled ( used for backup should primary device fail !)
You only need the network statement once, why are you creating multiple /30 network statements in ospf when you already have it declared with the 1 /8?. MAke sure you summarize routes on area border routers, to prevent flapping
I don’t know what you mean they are used for a backup.
The original configuration was done by a techie who is no longer available and I am trying to educate myself why certain configurations exist on this live network,
The /30 are used for OSPF interfaces network-type=point-to-point ?
The OSPF /30 listed are active in another device and should that fail I can enable them until the other device is operational ?
No that’s not how ospf is suppose to works. If your going to go in and enable and disable routes then you need to go back to static routes. Without seeing the big picture that looks to be someone learning how to use ospf. and was making a network statement for every network until they realized they only need to make 1 that summarizes them all.
You are correct that is not how OSPF is supposed to work! But when a Cisco certified network consultant first adds 10.0.0.0/8 in OSPF network and then proceeds to add /30’s within that ip range
with OSPF interfaces network-type=point-to-point, It had me wondering if this was best network configuration practice! maybe this type of configuration works OK on higher-end equipment but not so good on MIkrotik,
Yes I am trying to learn OSPF as the buck stops with me ?
Thankfully only the core routers - 2 X CCR1009’s which share the network load have these OSPF settings!
All other devices have interfaces using /30 like for example PTP ether interface /30, wlan interface /30 OSPF networks, OSPF Interfaces for both ether and wlan as network-type=broadcast
Your putting to much faith in a person.
ospf does not care whether you have 20 /30 routes or 1 /8 (or any other mask that summarizes the other ones) They do the same exact thing. I hate to tell you, but you can be cisco certified and still not know what your doing (especially at the ccna level), and you have not convinced me otherwise.
ospf is a open protocol there’s no such thing as it working different on different routers.
I understand this but while a 1 /8 could be used on the initial network configuration setup but once the 20 /30 etc have been setup then disable or remove the 1 /8 ,
We also have L2 switches used on the network, I wonder how much that adds a complication to the network design ?
Just to disambiguate a bit, I think you’re conflating two different things. But let’s talk certification for a minute
As someone pointed out
- You can be certified and still have no idea what you’re doing
- Certifications are the beginning of network engineering knowledge and are often broad in scope - they are nowhere near definitive
Be careful in using that as the only measure to justify a network design or practice
That said, i’ll share my relevant certs if it’s helpful - I have a background in both MikroTik and Cisco as I currently hold MIkroTik’s MTCINE expert internetworking certification and have been Cisco certified since 2003 - currently i’m an active Cisco CCNP ENT (converted from R&S)
Now for the difference between OSPF config and subnetting.
OSPF network statement configuration allows for the configuration of different mask lengths so that you can be more or less specific about which interfaces participate in OSPF and start exchanging hellos. It does not have any bearing on the subnet configured for an interface. You can specify a /32 length in OSPF for an interface that is configured as a /27 if you don’t want any other interface within a /24 that covers the /27 to be active in OSPF.
Conversely, you can be more permissive and configure a network statement as a /24 so that any /27 inside that range will automatically start sending hellos and participate in OSPF. It’s worth noting that the “network” style of OSPF config is older and most operating systems have been moving to interface level config for more than 10 years.
Now, to the point about subnet overlap in addressing - generally it’s not a good idea to configure overlapping subnets as it creates confusion and not all operating systems support this practice without the use of VRFs. A notable exception is during a network migration, if overlapping subnets exist between two different networks or multiple devices are being combined into a single device and that creates subnet overlap, sometimes it’s helpful to have that capability and live with the overlap for a short while until the network can be renumbered.
It’s also worth noting this is one of the great strengths of IPv6 - as a general rule, subnet overlap isn’t possible except in special ranges like doc prefixes or ULA.
It’s always nice to have someone chime in that can break a subject down to a level someone will understand. I only wish I could be so eloquent with my explanations.
Thanks for your contributions.
Thank you also for taking the time to give a very informative reply, however this raises two questions
(1) I would have thought that the certification examination would be extensive or it simply just know the answers to the questions that are “normally” asked from previous examinations!
(2) If certification cannot be trusted then feedback from customers who used your services should be posted on a open forum.