Enhanced VSI Protocol Capabilities


This section covers some further details and enhanced capabilities of the VSI protocol.

Slave Discovery

The VSI master continuously sends poll messages to all possible VSI slaves. The VSI message used in slave discovery polls is the switch get configuration command. When the VSI master receives a response to the discovery poll message or gets a Switch Get Configuration Trap from a VSI slave, it stops the discovery polling and starts the keepalive polling. It also performs a resynchronization. The keepalive and resynchronization functions are described in the following sections.

Master Keepalive Function

The keepalive is a simple request message that the master sends. It expects a response from the slave to detect when the slave cannot respond. The preferred VSI message for this function is the switch get configuration command. When the keepalive timeout is exceeded, the slave is declared unavailable, and subsequent connection messages for that slave are NACKed in the master.

To recover from a slave-unavailable condition, the master keeps sending keepalive messages, and the slave can also send traps to the master.

Database Synchronization

The master is responsible for keeping or reestablishing connection segment synchronization, in which database information is exchanged in order to enter a known state. In the event of a loss of communication between master and slave, synchronization is lost, but traffic is unaffected. After a loss of synchronization (for example, a LOS between controller and switch), there is an optional state of resynchronization isolation after which the resynchronization process starts.

Resynchronization isolation is an optimized method of identifying connections that need to be resynchronized. The master groups connections and associates a checksum with each group. When synchronization is lost, the slave is queried for the checksum associated with each group. Only the ones that do not match the calculated checksum are passed to the resynchronization process. The groups of connections are called checksum blocks.

Resynchronization can be triggered by a slave or controller rebuild or switchover, response timeouts, or new slave attachment.

Clock Synchronization

VSI provides messages so that a network clock application can select the clock source that the switch synchronizes itself and its interfaces to in order to synchronize the network to a single clock source. Such a network clock application is Network Clock Distribution Protocol (NCDP).

The switch advertises to the controller the clock quality, the interfaces that can be clock sources, and which interfaces can pass clocking information. External clock interfaces are also advertised as logical interfaces supporting no connections.

In a distributed slave model, the controller can send clock-related commands to any slave.

Flow and Congestion Control

VSI messages involve a transaction (and consequent transaction-id) for each message sent. As mentioned earlier, a response is expected for every command sent. As another flow-control mechanism, a windowing function prevents sending more messages than the number of outstanding (unacknowledged) messages subtracted from the transmit window. Flow-control windows in both the controller and the switch might be configurable.

In a distributed environment, when a VSI command is sent to a VSI slave, the VSI response is expected to be returned from the same slave to maintain the flow-control window properly.

An example of congestion control within VSI is the response code REJ (reject). It indicates that the VSI request was denied because of congestion control on the slave. The REJ response code has three possible reasons: master-detected VSI congestion, slave-detected VSI congestion, and slave-detected VSI congestion at another slave.

Message Passthrough

VSI provides a mechanism to pass messages transparently from the switch to the controller and vice versa. To that extent, four VSI messages are provided: the passthrough pass up command, the passthrough pass up response, the passthrough pass down command, and the passthrough pass down response (in which the direction up refers to the direction from switch to controller). The Passthrough Information Type parameter group contains the pg-format-id, the pg-length, a protocol-id, the info-length, and the info-data padded to be an integer number of bytes.

For example, in a PNNI application, the VSI slave can pass through ILMI messages for address registration to be processed by the controller. The slave can process ILMI keepalive messages in a distributed ILMI implementation, where the ILMI Managed Entity (IME) is in the controlled switch.




Cisco Multiservice Switching Networks
Cisco Multiservice Switching Networks
ISBN: 1587050684
EAN: 2147483647
Year: 2002
Pages: 149

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net