Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

Navigation

Understanding Edge Virtual Bridging for Use with VEPA Technology

Servers using virtual Ethernet port aggregator (VEPA) do not send packets directly from one virtual machine (VM) to another. Instead, the packets are sent to virtual bridges on an adjacent switch for processing. EX Series switches use edge virtual bridging (EVB) as a virtual bridge to return the packets on the same interface that delivered the packets.

What Is EVB?

EVB is a software capability on a switch running Junos OS that allows multiple virtual machines to communicate with each other and with external hosts in the Ethernet network environment.

What Is VEPA?

VEPA is a software capability on a server that collaborates with an adjacent, external switch to provide bridging support between multiple virtual machines and external networks. The VEPA collaborates with the adjacent switch by forwarding all VM-originated frames to the adjacent switch for frame processing and frame relay (including hairpin forwarding) and by steering and replicating frames received from the VEPA uplink to the appropriate destinations.

Why Use VEPA Instead of VEB?

Even though virtual machines are capable of sending packets directly to one another with a technology called virtual Ethernet bridging (VEB), you typically want to use physical switches for switching because VEB uses expensive server hardware to accomplish the task. Instead of using VEB, you can install VEPA on a server to offload switching functionality to an adjacent, less expensive physical switch. Additional advantages of using VEPA include:

  • VEPA reduces complexity and allows higher performance at the server.
  • VEPA takes advantage of the physical switch’s security and tracking features.
  • VEPA provides visibility of inter-virtual-machine traffic to network management tools designed for an adjacent bridge.
  • VEPA reduces the amount of network configuration required by server administrators, and as a consequence, reduces work for the network administrator.

How Does EVB Work?

EVB uses two protocols, Virtual Station Interface (VSI) Discovery and Configuration Protocol (VDP) and Edge Control Protocol (ECP), to program policies for each individual virtual switch instance—specifically, EVB maintains the following information for each VSI instance:

  • VLAN ID
  • VSI type
  • VSI type version
  • MAC address of the server

VDP is used by the VEPA server to propagate VSI information to the switch. This allows the switch to program policies on individual VSIs and supports virtual machine migration by implementing logic to preassociate a VSI with a particular interface.

ECP is a Link Layer Discovery Protocol (LLDP)-like transport layer that allows multiple upper layer protocols to send and receive protocol data units (PDUs). ECP improves upon LLDP by implementing sequencing, retransmission and an ack mechanism, while at the same time remaining lightweight enough to be implemented on a single-hop network. ECP is implemented in an EVB configuration when you configure LLDP on interfaces that you have configured for EVB. That is, you configure LLDP, not ECP.

How Do I Implement EVB?

You can configure EVB on a switch when that switch is adjacent to a server that includes VEPA technology. In general, this is what you do to implement EVB:

  • The network manager creates a set of VSI types. Each VSI type is represented by a VSI type ID and a VSI version--the network manager can deploy one or more VSI versions at any given time.
  • The VM manager configures VSI (which is a virtual station interface for a VM that is represented by a MAC address and VLAN ID pair) . To accomplish this, the VM manager queries available VSI type IDs (VTIDs) and creates a VSI instance consisting of a VSI Instance ID and the chosen VTID. This instance is known as VTDB and contains a VSI manager ID, a VSI type ID, a VSI version, and a VSI instance ID.

Published: 2014-04-23