Hot answers tagged

7

A router doesn't "change data-link header". It throws away an encapsulating frame on reception and creates a new encapsulating frame for transmission. A router forwards between networks. These networks use a certain data link layer (L2) protocol and that is employed by the router to reach the next hop. For that, it encapsulates the packet to be ...


4

There's not really a hierarchy. It's really more of a web. There are national (or international) ISPs that connect to a large number of other ISPs. There are smaller ISPs that connect to only a few other ISPs. Then there are large companies that have their own backbone (Google, Microsoft, Amazon, Facebook, etc). There's no "rule" about who can ...


3

When we say that a router "operates at the network layer," we mean that its main function -- routing packets -- is based on network layer information (i.e. IP addresses). But in fact, all IP devices use all layers (in the TCP/IP model) in order to communicate with other devices.


3

A route: object should include an origin: AS<x> attribute associating the route with its BGP origin AS. You can create multiple route objects to announce the same prefix from multiple ASNs. An as-set: has a different use. If you have downstream ASNs, you may list them all in an as-set: then tell your transit providers & peers about your as-set. ...


3

You're creating an asymmetric routing situation. The destination may be different, but the source will be the same, so return traffic will always flow across the same ("default") tunnel. One way to get around this is to do "twice nat"; rewrite source and destination. This doesn't necessarily have to happen on the same ASA, but often makes ...


2

(from comment) The overhead is only visible (and countable) outside the tunnel. Roughly you should see (outer packet header + GRE header) * number of packets more traffic on the outside (physical) interface than on the tunnel interface. Depending on where you're watching, a physical interface might also be counting L2 traffic (add the L2 header size * number ...


1

Add no auto-summary to your BGP configurations


1

A router's function is simply to accept a data packet from one iterface and send it out of another one, hopefully bringing the packet one hop closer to its destination. The router uses a routing table to store selected routes to destination networks, and a forwarding table in hardware with the exact information needed to route a packet. Each route in the ...


1

Yes. An increase in path cost only means that the path may be less preferred than another (or it may be the only path). If the route is still in the routing table, the router will still forward the packet.


1

can we still transfer packets between the two routers? Yes. But your network might not work efficiently or at all. Basically, most routing algorithms in practice use Dijkstra's algorithm (OSPF, IS-IS, SPB) which is faster to converge. Any router can only use the information that is has gained access to - effectively, until the network has converged, there's ...


1

Do yourself a favor and don't use NAT where it's not required. You're tunneling between private address areas and there's no address collision, whence no need to complicate things with using NAT. Once you're routing transparently, you've simply got two paths between the networks. I'd use policy routing with fall back. ECMP requires session tracking and ...


1

(This will tread on "historical trivia") The tiered construct was the original vision of the ARPANet (what became the internet.) The modern internet does not work like that. In fact, the early internet -- and ARPANet before it -- didn't entirely work like that either. With the advent of BGP, any network could be anywhere, connected to anyone. ...


Only top voted, non community-wiki answers of a minimum length are eligible