Star, Hub, Mesh, Bus, Ring

Star Topology

• Devices communicate through a central device
  • Hub
  • Switch
• Cable issues only impact affected device
  • Having multiple ports gives us multiple options. If you've only got the one link, if it goes away, that's your single point of failure.
• Requires more cabling
  • If you want to add another computer to a star network, the cable has to go from the physical switch all the way up to the new workstation.
• Commonly used in Ethernet networks
• Easy to troubleshoot

Hub-and-Spoke Topology

In a Hub and Spoke topology, a central device (the "hub") connects to multiple peripheral devices (the "spokes"). All communication between the spokes must pass through the hub. This topology is often used in networking, transportation, and telecommunications systems.

• Workstations connected to switches

• When we draw connections between the routers, many times if it's a straight line, it's assumed to be Ethernet

• If the link it's represented with a lighting bolt it's assumed to be a serial or a WAN link.

  • Probably representing connection between two far places or even different cities or states.

• The problem is that if the central router goes down, all of the other branch sites are offline.

  • See it as if the central router is the headquarters, and for other routers to communicate the traffic has to come to that central router, the headquarters, and then go back out to its destination.

• Isn't always an optimal design since every operation has to go through a single router, but sometimes we do this because of cost savings, we are not buying individual dedicated links between all of your branch offices.

• Locations connect through a central site

• Less optimal routes

  • Adding an additional point of failure.

• Common with WAN links

  • Sometimes used with VPNs

• Recurring fee for each link

  • Not true if we use IPsec VPN

• Helps minimize expenses

Key Differences between Hub and Spoke and Start topologies:

1. Communication Paths:
   • Hub and Spoke: Spokes only communicate with each other through the hub.
   • Star: Each device communicates through the central device, but they don't communicate with each other directly.
2. Use Cases:
   • Hub and Spoke: Typically used in larger networks or WANs, where central control and traffic management are needed (e.g., for VPNs or internet backbones).
   • Star: Common in LANs, with a central switch or router connecting devices in an office or home network.
3. Redundancy:
   • Hub and Spoke: If the central hub fails, the entire communication network fails.
   • Star: Failure of the central hub causes communication failure, but individual devices' failure doesn't affect others.

Full-Mesh Topology

• Full-Mesh = Everything is connected to everything.

  • Every single device connected to every single other device
    • Not very scalable, it's gonna get expensive.

• Partial-Mesh = Multiple hubs (dual or more hub design)

  • Each router has a link to hub1 and hub2 (dual hub).
  • If one central site goes down, we'll take a less optimal route, but everything is still working.

• Locations connect through a central site

• Less optimal routes

• Common with WAN links

• Recurring fee for each link

  • This is gonna be more expensive but we are connecting everything to everything
  • Every time you add a new site, all other sites have to have an additional link added to them
  • Not very scalable because of price

• Helps minimize expenses related to Outage and resiliency but the monthly normal expenses would probably be higher.

Bus Topology

• All devices connect to one cable
• Media is shared/contested by all other devices
  • Anytime somebody wants to talk first they have to look and see if the space is available: "is the coast clear?" and the jump.
  • If right at the same time two host jump now you have a collision, now you've got to back off and try it again.
  • It's an old technology but there are still networks that work like this, for example a CAN bus means to share media for everybody on campus.
• Older Ethernet networks
• Single network segment
  • You do not need a hub, no need a switch, no need of individual runs of cable going from each computer to a central site, just from one computer to the next.
• Same collision domain
  • Analogy: "Two people trying to have a conversation on a telephone. It's silent, I'm waiting for you to say something, you are waiting for me to say something and then suddenly we both speak at the same time, that's a collision"
  • Carrier sense multiple access with collision detection, it's called CSMA/CD, is what we can use to look if other people are speaking at the moment.
  • Every environment that uses CSMA/CD has a potential for collisions, which is just inefficient with sending traffic. It means you have to stop and do it all over again.
• Logical vs physical bus topology
• Inexpensive
  • No need any infrastructure
• Cable issues can affect entire topology
  • With British Naval Connectors (BNCs) you typically use a T and the T would bring in the network from the other computer that already had it, it would drop it off to your computer, and then on the other side of your computer that is where we go to the next computer. If there is not a computer on this side we would put a device called a terminator (Doesn't let the network leak out) - screw and cap.
  • If we want to add another computer we unscrew the cap, move a wire over a T connector and just put that terminator on the end. It was simple.
  • The problem is that it was shared and it wasn't very fast. Star networks have taken over first with the hub, then with the switch.
• Not commonly used

CSMA/CD vs. CSMA/CA

• CSMA/CD (Carrier Sense Multiple Access with Collision Detection) and CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) are protocols that manage how devices share a communication channel in a network.
  • CSMA/CD is primarily used in wired networks like Ethernet, while CSMA/CA is commonly used in wireless networks, such as Wi-Fi.
• The main difference lies in their approach to handling collisions:
  • CSMA/CD detects and resolves collisions after they occur, whereas
  • CSMA/CA attempts to avoid collisions before they happen.

Ring Topology

• A ring topology is a local area network (LAN) in which the nodes (workstations or other devices) are connected in a closed-loop configuration.
• Ring topologies aren't used heavily in local area networks anymore, but they are still commonly found in wide area network connections as an FDDI ring.
  • An FDDI ring is a Fiber Distributed Data Interface ring, which allows for a network that can communicate up to 120 miles in range, uses a ring-based token network as its basis, and uses two counter-rotating token ring topologies to comprise the single network.
  • This provides redundancy for the network because if one cable is broken or fails, the other can maintain the network operations. The token is used to control which device can communicate on the network, preventing congestion or collisions.

Physical Bus Technology

• (BNC)
  • Not very fast, antiquated
• Coaxial cabling
  • Thicknet
    • 1st version of this is called thicknet, and you would actually connect to with something called a vampire tap that would clip onto the outside.
  • Thinnet
    • The thinnet version worked with the terminators
• Network taps
  • Vampire tap
    • Thicker cables running across the back of the wall almost like a graden hose with tiny holes to deliver water to each plant but for each PC.
  • T-Connector
• Serial cabling
  • Not super common but it's a possibility.
  • Going from a serial port
    • USB = Universal Serial Bus
    • A lot of this today has pivoted to using USB instead and in different formats of USB not so much for networking but just for general device communication.

Spine and Leaf Network Architecture (for inside servers)

Three-tier (for end-points)

• Core layer
  • Focuses on high-speed packet switching
• Distribution layer
  • Responsible for routing and filtering traffic between different network segments, ensuring efficient communication and controlling access to resources
• Access layer
  • Primarily deals with end-user connectivity