Another topic of discussion at the ongoing OEM Windows Server 2012 R2 events taking place around Australia is around when and where you should use NIC teaming versus leveraging SMB multichannel capabilities, so today I will walk through a few scenarios of most interest to SMBs and highlight a few resources.
With official support for NIC teaming introduced in Windows Server 2012, Microsoft delivered what many partners and customers had been requesting for a long time, ownership of NIC teaming capabilities without introducing third party solutions. The main areas of NIC teaming usage in small business scenarios are as follows…
Physical NIC Aggregation
NIC aggregation means that all workloads that can benefit from failover, bandwidth increases, load balancing etc can benefit from a NIC team. With support for up to 32 network adapters per team, we have a great deal of flexibility. This will help with most types of network traffic, including file, web and more.
Virtual Switch Aggregation
We can now assign a team to a Hyper-V virtual switch, so that all of the VMs that leverage that switch can share the aggregated bandwidth. Previously you may have provided access to a single physical NIC per VM, in many cases a gigabit per NIC, but now, for example, you could team 4 gigabit NICs into a 4 gigabit aggregated connection, assign it to a Hyper-V virtual switch, and now make assign that virtual switch to a group of Hyper-V VMs so that they can use what they need rather than being capped with a single NIC, and potentially leaving bandwidth underutilised in some of the virtual swtiches.
SMB Bandwidth Aggregation
If the purpose of the additional network adapters is to take advantage of higher SMB traffic throughput, SMB multichannel can do this automatically for you. Just remember that other types of network traffic, such as FTP, HTTP, HTTPS etc won't benefit from the bandwidth aggregation. This is an approach you should consider when looking at centralising the storage of VHD or VHDX files centrally, SQL workloads, or when high speed files copies are needed between servers.
End to end failure detection is implemented, allowing data to continue flowing even if there is a NIC, cable or switch failure.
Multiple network paths can be detected and utilised automatically, allowing for failover to take over alternate paths when needed.