Friday, January 10, 2014

Virtualization Networking Basics Part 1

To start off the New Year, I wanted to go over some networking basics for virtualization. This will be a multi-part series. First, lets go over some of the key networking terms.

Network Definitions
  • Network: Physical connection that allows computers to communicate
  • Packet: Unit of transfer, bits carried over the network
  • Protocol: Agreement between two parties as to how information is to be transmitted
  • Broadcast Network: Shared communication medium
  • Arbitration: Act of negotiating the use of a shared medium
  • Point-To-Point Network: A network in which every physical wire is connected to only two computers
  • Switch: A bridge that transforms a shared-bus (broadcast) configuration into a point-to-point network
  • Router: A device that acts as a junction between two networks to transfer data packets among them
  • Gateway: A device that connects two networks communicating over different protocols
Next, lets take a look at the OSI structure and the associated devices for the network connection.

The OSI layers start at the application layer, presentation layer, and session layer which are generally considered as one block from a networking standpoint; followed by transport layer, network layer, datalink layer, and physical layer. You can think of the NIC and operating system at the physical layer, datalink layer, network layer, and transport layer; and everything above the operating system in the session layer, presentation layer, and application layer.

Lets regard the left side of the diagram as the first ESXi host and the right side as the second ESXi host, all the traffic would pass through the OSI layers sequentially starting with the application layer in the first host down to the physical layer and then up from the physical layer to the application layer in the second host.

Ethernet is a OSI layer 2 datalink technology, Ethernet is now the standard for most data center and campus communications. Ethernet is defined by a packet or frame. Each frame contains a destination and source address, a MAC addresses is 48-bits. The type defines if data portion of the Ethernet frame is an IPv4 datagram (0x0800) or an IPv6 datagram (0x86DD) datagram. The data part of the layer 2 frame contains the layer 3 datagram.

Now lets talk a little bit about the switch itself. The switch is an entity that sends traffic from host A to host B. Its key function is to create a forwarding table and learn the location of each node by looking at the source address of each incoming frame. It then forwards each incoming frame to the port where the destination node is located. This helps to reduce the collision domain, makes more efficient use of the wire through Full Duplex, and the ESXi hosts do not waste time checking frames that are not destined for them like in the legacy Ethernet arbitration model.

Switches will broadcast traffic to all the ports in a given subnet if the destination address is unknown and then learns from the responses. One thing to keep in mind, switches will not reduce the broadcast domain it reduces the collision domain. Most switches provide very sophisticated features like keeping track of VLANS, applying policies by access control lists, and providing quality of services (QOS) for granular link level traffic control. In a data center environment a layer of switches is a switch fabric.

When you compare a switch with a router, the fundamental concept is the same; it moves packets from one host to another. A router looks at the IP packet destination and checks its routing table to decide where to forward the packet. A router is a termination point for a particular subnet, all broadcast traffic that is generated in a subnet is not passed on by the router. Unlike a switch, when a broadcast frame is seen by the router, it will not send out the broadcast frame to the other subnet. Instead it is going to terminate the broadcast. This becomes really important when trying to design hierarchical and scalable network designs.

This is a quick look of a traffic domain and boundaries.

In the terms of the link layer technologies, there are a number of network interface layers available including Fast Ethernet, Gigabit Ethernet, Fibre Channel, and Wireless LAN. Gigabit Ethernet is the most popular with speeds ranging from 1 Gbps to 40 Gbps. Also, there are a variety of transmission modes offered as well, such as twisted pair and optical fiber. The correct transceiver needs to be matched for the corresponding transmission mode. When designing a network infrastructure, a substantial amount of the cost is incurred with the types of cabling that is used to build out the infrastructure. That is why virtual networking is becoming popular, not as much expense is required in cabling the data center when using a virtual network infrastructure.

News: Top vBlog 2016 Trending: DRS Advanced Settings