Why is NIC Teaming Dynamic mode better than Address Hash for non-Hyper-V servers?

Why is NIC Teaming Dynamic mode better than Address Hash for non-Hyper-V servers?

Q. Why is NIC Teaming Dynamic mode better than Address Hash for non-Hyper-V servers?

A. The Dynamic mode was introduced as a new load balancing option in Windows Server 2012 R2. Dynamic took the best of address hash and Hyper-V port and should always be used in Windows Server 2012 R2.

The reason it is better than Address Hash for non-Hyper-V traffic is due to the introduction of flowlets with dynamic mode. The is best understood by looking at how load distribution works between the two modes.

With Address Hash, each time a new flow is detected, the new flow is assigned statically to a team member. The assignment is not based on existing traffic on any of the members of the team. Once assigned, a flow will never move between members of the team. So it's possible for several very large flows to all be on the same team member, while other team members have little traffic. This may result in delayed or dropped packets for these over utilized members.

On the other hand, dynamic distribution constantly watches flows and any time there is a pause in the flow, when the flow resumes the traffic on members is evaluated and the flow moved to the members with less traffic. This means traffic is constantly rebalanced avoiding any one member having significantly more than others. Each of these flows is known as a flowlet.

It is this flowlet rebalancing that makes dynamic always better than address hash.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish