Network Flashcards
(16 cards)
Distributed Routing
Definition: The routing process is handled by multiple nodes (routers, switches, etc.) working together, but they may still rely on some centralized control or coordination.
Key Traits:
Nodes share routing information (e.g., via protocols like OSPF, BGP).
Decisions are made locally, but often based on a global (or semi-global) view of the network.
Can still have hierarchical or partially centralized elements (e.g., route servers in BGP).
Example: The Internet’s routing system (BGP) is distributed but not fully decentralized because it depends on trusted entities like ISPs and route authorities.
Multiple nodes cooperate, but some centralization or hierarchy may exist.
Decentralized Routing
Definition: No central authority or hierarchy exists; routing decisions are made entirely in a peer-to-peer (P2P) manner.
Key Traits:
No single point of control or failure.
Nodes independently determine routes (e.g., via gossip protocols, DHTs, or mesh networks).
Often used in blockchain networks, darknets (e.g., Tor), or ad-hoc wireless networks.
Example: Bitcoin’s node communication or Freenet’s routing is decentralized.
Fully peer-to-peer, no central coordination.
Centralized Routing
Centralized Routing means that a single (or a few) controlling entity makes all routing decisions for the network.
Key Traits of Centralized Routing
Single Point of Control: One central node (or a master controller) computes and dictates routes for all other nodes.
Dependent on Central Authority: If the central node fails, the whole routing system may collapse.
Easier to Manage: Simpler to optimize routes since one entity has full network visibility.
Less Scalable: Bottlenecks can form as the network grows.
Examples of Centralized Routing
Traditional SDN (Software-Defined Networking): A central controller (e.g., OpenFlow) decides paths for all switches.
Legacy WAN Routing: Some older networks relied on a primary router to dictate paths.
Air Traffic Control Systems: A central system directs flight paths.
One boss decides routes.
Is Centralized Routing the Same as Global Routing?
No, but they can overlap:
Centralized Routing = Control is in one place (single decision-maker).
Global Routing = Routing decisions are made with full network knowledge (but not necessarily centralized).
How They Relate:
A centralized system often uses global routing because the central node has complete topology info.
But global routing can also exist in distributed systems (e.g., link-state protocols like OSPF, where each node has a full map of the network).
Example:
Centralized + Global: SDN controller computes optimal paths for all switches.
Distributed + Global: OSPF routers each have a full topology database but compute routes independently.
Global
Routing uses full network knowledge (can be centralized or distributed).
Another word for Subnet Address
Network ID
Host bits calculation
32 - CIDR number
What does the CIDR number represent?
Network bits
How do you find the # of usable hosts (aka usable IPs) (how many devices can you have in your network)?
32 - CIDR = x
2(^x)-2 = #
-2 (for network ID, and broadcast address)
How do you find subnet address?
If first subnet, it’s just the Address Space number.
The following subnet is simply previous subnet Broadcast Address + 1
How do you find the Broadcast Address?
Replace host bits by 1
How do you determine the range?
Subnet Address + 1 to Broadcast Address - 1
Static Port Forwarding (a NAT traversal solution)
Manually map a port to an internal IP.
Best for: Hosting servers (web/game).
Pros: Reliable, no middleman.
Cons: Manual setup, security risks.
(as for static port forwarding, you’re just manually whitelisting some specific port number for incoming public traffic)
Universal Plug and Play (a NAT traversal solution)
Apps automatically open ports.
Best for: Gaming, P2P apps.
Pros: No user setup needed.
Cons: Security risk, not always available.
(When you’re using a device or application (e.g., a game console, torrent client, or VoIP app) on your private network, UPnP allows that app to automatically request your router to temporarily open a specific port for incoming public traffic—but only while the app is actively using it.)
Relaying (TURN server) (a NAT traversal solution)
Data goes through a middleman server.
Best for: VoIP (Zoom), strict NATs.
Pros: always work
Cons: Latency (overhead) due to middleman, costs money
(so for relay, it has a specific server for public data, and when that server has public data, it relays back to the private network, and that private network has authorized that specific server specialized for handing public data?)
NAT traversal failure
NAT traversal failure happens when devices behind a NAT (like routers in home or office networks) can’t establish a direct connection with external devices due to strict firewall rules, incompatible NAT types (like symmetric NAT), or misconfigured settings. Techniques like STUN, TURN, or ICE may fail if the NAT blocks incoming connections or lacks proper port forwarding, forcing data to route through slower relay servers or causing connectivity issues entirely. This often disrupts real-time applications like VoIP calls, online gaming, or video conferencing, leading to lag, dropped connections, or complete communication failures.
NAT policies and traversal failures prevent unsolicited incoming connections.
No Port Mapping: NAT doesn’t automatically preserve a reverse path for incoming traffic unless the private device first “punches a hole” (e.g., via UPnP, manual port forwarding, or protocols like STUN/ICE).
