NIC Teams are introduced with descriptions of the three load balancing modes and scenarios where each is useful in creating a high-performance network.
- [Narrator] I'm going to kick off this course with a little discussion on high-performance networks and I'd like to start that by making sure we're on the same page as to what high-performance means. When I've taught beginning network students the concept of high-performance, I've usually said that high-performance means speed. That is a bit oversimplified though, and in this segment on NIC Teaming, we'll see an example of why that is. See, NIC Teaming is all about using multiple network adapters to improve performance in and out of a server.
One of the most common and most misinterpreted measures of network speed is the time it takes to download a large file. I can save you some time up front and tell you that if you want to download a single file faster, just buy a faster adapter. If, on the other hand, high-performance means your server is handling a lot of network requests and you want to distribute that load to speed up the entire network, NIC Teaming is a good candidate for your needs.
The basic concept of NIC Teaming is taking multiple physical adapters and making them appear to Windows to be a single interface. What this does for us is it allows us to assign a single IP address or a single identity to the server, and then take that IP address and apply it to multiple physical adapters and take advantage of the higher bandwidth. The higher bandwidth comes from having multiple pipes that share the load and allow multiple streams of data at the same time.
And each of these separate streams can claim one pipe for themselves, which makes the communication more reliable and fewer lost packets. Less packet loss means fewer retransmissions, which also adds to higher performance. Now, NIC Teaming can be done a few different ways, and the different approaches to NIC Teaming can be described by both the teaming mode and the load balancing mode. The load balancing mode serves different types of network traffic scenarios.
For our first scenario, let's look at one that's been around for a long time, the basic file server. A file server could easily be expected to make thousands of files available to hundreds of users. Address Hashing is a load balancing mode designed specifically for this type of environment. It uses multiple NICs that appear as one to both Windows and to the network to distribute this load. It sets one NIC in the team aside to accept inbound requests, that way everyone asking for their files can have their requests queued and prioritized.
Downloads are then assigned to the remaining interfaces on a first come, first serve basis. Instead of a single adapter slipping all frames on the network in whatever order they're created, one pipe delivers one message and then becomes available for the next. Another load balancing mode is the Hyper-V Mode, named for the type of traffic that it was designed to balance. If you have a Hyper-V server, hosting three virtual machines, you have three servers worth of traffic coming into and out of your physical host server.
Now I mentioned before that multiple NICs don't make one download that much faster. Similarly, one really fast network adapter is not likely to be the best solution for multiple virtual server data streams. The Hyper-V load balancing mode allocates physical pipes to each virtual machine. This allows each virtual server to have more reliable communications with the network. And there's one other load balancing mode that was introduced in Windows Server 2012 R2 called Dynamic Mode.
This mode allows the operating system to decide between Hyper-V and address hashing and it can change between the two on the fly. In the next segment, we're going to take a look at another way to differentiate different types of NIC teams based on the switch or switches that the physical adapters connect to and what role those switches will play.
- Configuring a network interface controller team
- Switching embedded teaming
- Remote enabled direct memory access NICs
- Configuring virtual machine queue
- Enabling and configuring SR-IOV
- Understanding software-defined networks (SDN)
- Reviewing SDN network requirements and deployment scenarios