As was discussed in the last newsletter, all of data center network design guidelines and technologies that IT organizations have used for the last decade are now being questioned. The good news is that IT organizations have a lot of new and emerging technology and design options that they can implement. The bad news is that the breadth of available options can be overwhelming. We will use this newsletter to describe a series of discussions on data center networking that we are driving the goal of which is to help IT organization evaluate their options.
Paving the way for the flat network
As mentioned in the last newsletter, in the traditional data center physical servers were typically dedicated to a single application. Therefore, network policies could be implemented by configuring the physical port of the network access switch with the Access Control Lists (ACL), Quality of Service (QoS) classifications, and the rate-limiting characteristics required for specific applications.
Once data center servers are virtualized, multiple disparate applications can be running on virtual machines (VM) sharing a single physical server. IT organizations need the ability to manage network traffic using the same granular policies relating to security, QoS, and regulatory compliance for applications running on VMs as it did previously for applications running on dedicated physical servers.
One way that this can be accomplished is if the network access switch is capable of mapping each VM to a separate, individually configurable virtual switch port. The network access switch management system can then maintain a profile for each VM, including the VM’s IP address and virtual MAC address, VLAN memberships, ACLs and QoS classification.
As was also mentioned in the last newsletter, the spanning tree protocol has been a mainstay of data center networks because it eliminates loops. Unfortunately, the spanning tree protocol prevents all available forwarding resources in a redundant network design from being simultaneously utilized. If this constraint can be eliminated, then the bandwidth efficiency data center networks with redundant links can be greatly improved by assuring that the parallel links from the servers to the access layer and from the access layer to the core layer are always in an active-active forwarding state.
In order to help IT organizations evaluate the existing and emerging data center designs and technologies, we will be driving a series of six monthly discussions on topics related to data center networking. Joining us in these discussions are six of the leading vendors: Arista, Avaya, Brocade, Cisco, Extreme and HP. Each of the six will respond to the questions we ask and will comment on the responses of the other sponsors. Each vendor’s response and comment will focus on technology and design issues. The vendors will not, for example, discuss products or other vendors.
The first discussion focuses on alternative to the spanning tree protocol. That discussion goes live on Aug. 4 and you can read it here.
