| 1. | Where is the most appropriate place to connect a block of enterprise (internal) servers? Why? |
| Answer: | You should connect a block of enterprise servers into the core, just as you would switch blocks. The server farm building block should have a layer of access and distribution switches, just as in any switch block. This maximizes connectivity from the servers to all other devices in the network. In effect, all users will see the same number of switch "hops" to access a server. Connecting into the core also provides maximum scalability because you can add more server blocks in the future. |
| 2. | How can you provide redundancy at the switch and core block layers? (Consider physical means, as well as functional methods using protocols, algorithms, and so on.) |
| Answer: | In a switch block, you can provide redundancy through two distribution switches. Each access switch can be linked to both distribution switches for fault tolerance. The Layer 3 distribution layer allows both uplinks to be used at the same time, with little or no failover time required. In the core layer, a dual core can be used with two core switches. Each distribution switch has dual uplinks, with one link to each core switch. Here, the redundant links can stay active for load sharing and redundancy, thanks to the Layer 3 routing protocols running in the distribution and core layers. |
| 3. | What factors should you consider when sizing a switch block? |
| Answer: | Consider traffic types, flows, and patterns, as well as the size and number of common workgroups. Additionally, the Layer 3 switching capacity in the distribution layer should be sized according to the amount of traffic crossing from one VLAN to another. |
| 4. | What are the signs of an oversized switch block? |
| Answer: | The distribution switches begin to become bottlenecks in handling the interVLAN traffic volume. Access list processing in the distribution layer also can become a rate-limiting factor. Broadcast and multicast traffic forwarding can slow the Layer 2 and Layer 3 switches in the block. |
| 5. | What are the attributes and issues of having a collapsed core block? |
| Answer: | Attributes: Cost savings (no separate high-end core switches) and design simplicity. Issues: Scalability becomes limited. |
| 6. | How many switches are sufficient in a core block design? |
| Answer: | Two switches are usually sufficient in a core block, offering load sharing and redundancy. However, you can add core switches as the size of the network and core traffic flow dictates. |
| 7. | What building blocks are used to build a scalable campus network? |
| Answer: | The switch block is the template used to group access-layer switches and their respective distribution-layer switches. Switch blocks then are connected into the core block to build a scalable network. Depending on the other resources that are present in a campus network, other building blocks can include a server farm block, a network management block, and an enterprise edge block. The service provider edge block exists outside the campus network, although it does interface with the enterprise edge block. |
| 8. | What are two types of core, or backbone, designs? |
| Answer: | Collapsed core and dual core. |
| 9. | Why should links and services provided to remote sites be grouped in a distinct building block? |
| Answer: | Remote sites and roaming VPN users should be considered corporate users, as if they were connected directly inside the enterprise network. These users should enjoy the same efficient access to any enterprise resource that internal users have. Additionally, VPN tunnels should terminate in a secure area of the enterprise network. Connections into the Internet (through service providers) are just the inverseusers from all parts of the enterprise need equal and efficient access to resources located out on the Internet. Therefore, a separate building block connected into the core is justified. |
| 10. | Why should network-management applications and servers be placed in a distinct building block? |
| Answer: | Network-management applications must be capable of polling, querying, or accessing devices anywhere in the campus network. Moving these resources into a distinct building block provides redundant and efficient access into the network core so that all devices can be managed with equal access. |
