CBT Nuggets - Switch 642-813
Video 1 - Intro to Switch
Video 2 - The Switches Domain - Core Concepts and Design
Hub
Every frame/packet is flooded out of all ports (if get to 20 or so daisy chained) you get alot of collisions and things sloooooow down
Layer 2 switch
Every port is a collision domain, your traffic only goes to the destination port, unless it is a broadcast, that was a limiting factor .... so its a scalability limitation, whereby if the devices get to many broadcasts on a segment(but that would have to be lots) it would affect performance
Then we saw that we could scale things up with VLANS, they changed everything, we didnt have to have routers everywhere to segment the network, this is where we used "router on a stick" for inter-vlanning, but this was a limitation tooo as the trunk link from the router to the switch was the bottleneck.
Layer 3 Switch
Then L3 Switches came out and changed it, it splices a router into a switch.
Here we can have SVI (Switch Virtual Interfaces) as the VLANs and the bottle neck now is the backplane speed of the switch (typical lots though, Gigs)
Now this is where design comes in ... Where do the L3 switches go etc
This is where the Enterprise Composite Model (ECM) comes in, however before that:
Typically the first basic network design consists of a L3 switch and x2 L2 switches, but already this is way to much for a small/home business,
Performance
The ECM addresses performance by dividing functional areas into modules, and connecting them together over a high speed backbone. This allows for efficient summarization of networks, and more efficient use of high speed uplink ports.
Scalability
With its modular approach, the ECM is scalable. Simply add on more function modules as required.
Availability
Again, modules can be connected in a redundant fashion to the core/distribution layers.
Role of switches in ECM
Access – layer 2 switches connecting workstations and other devices to the distribution. Here, basic protocol filtering and QoS marking would happen
Distribution – either L2 or L3 switches depending on the size and needs. These switches will bring in multiple access switches, perform more advanced QoS and filtering. Distribution connects back into the core/campus backbone
Backbone – again, either L2 or L3 switches depending on needs. Primary goal is to switch packets, so switches will be acting on tagged QoS packets rather than marking, and perform basic routing.
A block doesn't represent just 1 vlan, it could be/typically lots of vlans
vlans provide scaliability and don't slow you down (as in the old days any time you wanted to segment you had to have a router - which are slow, compared to switches.)
With the L3 switches we now have CEF (Cisco Express Forwarding) for caching info and route between vlans as if we didnt have them (so the number of VLANs is not an issue, it is not a limitation.)
Cisco are trying to move people in the direction of layer 3 vlans /localised vlans
vlans stay within there switch blocks - very efficient
The Managment vlan (as covered in the CCNA-S)
is where that port on the device is for the ssh, tftp upgrades, syslog all traffic to and from device for management (Out Of Band) - isolated, reason doesnt always work like this, fairly recent kit to implement/wealthy companies etc
