Survivability Overview

Survivability is a collection of features that enables the system to extend VoIP service availability. These features include support for redundant SIP soft switches and local call control in the event of WAN link failure, softswitch failure, or during periods of network congestion that result in loss of connectivity.

Dynamic WAN links (DHCP and PPPoE, for example) can renegotiate their public IP address at any time, interrupting VoIP services. All system services must be restarted when the WAN connection is restarted, because the WAN IP address may have changed. Because Survivability must communicate with the softswitch/IP PBX, Survivability is restarted when WAN link renegotiation occurs, interrupting any local calls that are in progress.

The below given matrix specifies the supported operation modes and SIP transport types.

• Availability—Detects when the softswitch is unreachable.
• Redundancy—Switches to a redundant softswitch:
  • Makes and receives station-to-station calls while the softswitch is unreachable
  • Makes and receives PSTN calls while the softswitch is unreachable.

To switch to Survivability mode, the EdgeMarc must be able to detect when the softswitch is unreachable. The softswitch may be unreachable for the following reasons:

• The softswitch is down.
• A router on the path to the softswitch is malfunctioning.
• The network is physically disconnected.

The most reliable way to ensure that the softswitch is reachable and available is to make a request at the application layer. If the request receives a response, the softswitch is reachable at the IP layer and is servicing requests.

How Survivability Works

SIP requests sent to and received from the softswitch are monitored for softswitch availability. If requests are sent to the softswitch but no responses are received in a configured time interval, the softswitch is considered unreachable and a backup server (if available) is selected as the current server. The default time interval is configured so that failover occurs before the phone has finished re-sending its request. This enables the call to be connected with only a slight delay.

A configuration option also allows the active softswitch to be monitored with keepalive messages. The messages are sent at a configurable intervals with time allowed for a response to be received. If too many messages are unanswered, the softswitch is considered unreachable.

When a softswitch is marked as unreachable, the EdgeMarc uses a different keepalive mechanism to determine when it becomes available again. The EdgeMarc sends keepalive messages to failed servers using a backoff algorithm that progressively increases the interval, until a maximum is reached. By default, the maximum interval is longer than the one used when no other backup server is available.

An upstream EdgeMarc or EdgeProtect device can send a specially marked keepalive response to inform the downstream EdgeMarc about loss of connectivity to the softswitch. This causes an immediate fallback to the backup server or Survivability mode in the downstream EdgeMarc without a time delay.

The EdgeMarc can choose the active softswitch from a list of multiple redundant softswitches based on priority and availability. It can obtain the list of redundant softswitches dynamically by doing a DNS SRV lookup on the SIP Server domain name or from a list entered in the appliance user interface. When multiple IP addresses are configured, the highest priority one is used. If this server becomes unreachable, the next reachable server in the order of priority is used. If a previously unreachable server becomes available and it has a higher priority than the currently used server, the higher priority server is used again.

If no softswitch is available, the EdgeMarc enters Survivability mode and handles call signaling itself. Because the EdgeMarc forwards all messages between the phones and the softswitch, it knows the address of all phones and can direct calls to the phones itself.

A SIP PSTN gateway can be installed on the LAN side of the EdgeMarc and used for inbound and outbound calling during Survivability. (Normally, the gateway functionality is provided by the softswitch.) The gateway can be configured to send any incoming calls to the EdgeMarc, and the EdgeMarc can be configured to use the gateway as the default destination for calls not directed to another local phone.

When connectivity to the softswitch is restored, the EdgeMarc automatically returns control of all subsequent call requests to the softswitch. Calls in progress that were established while the EdgeMarc was in fallback mode are not disrupted.

Survivability in Transparent Mode

The EdgeMarc now supports Survivability when the transparent option is selected for the Session Initiation Protocol (SIP) mode.

An EdgeMarc is typically deployed with two separate IP addresses:

• Public IP address used for interactions with the WAN
• Private IP address used for communications with locally-installed phones and other devices

The devices attached to the EdgeMarc communicate with the appliance using the private IP address, and the EdgeMarc communicates with the WAN using the public IP address. The network softswitch that provides feature services knows the public IP address but has no knowledge of the private address. The treatment of traffic sent between phones and the network depends on the choice of SIP mode.

In normal (default) mode, a phone connected to the EdgeMarc is configured with a private IP address and views the appliance as if the appliance is a softswitch. The Layer 3 signaling between the phone and the EdgeMarc is terminated at the appliance, and a separate Layer 3 signaling link is established between the EdgeMarc and the softswitch on the network. The phone device cannot see all the way to the network softswitch, and the softswitch cannot see all the way to the phone.

In multi-homed outbound proxy mode, the EdgeMarc functions as a proxy server. Each phone connected to the appliance is configured with the IP address of the network softswitch. An outbound proxy address is also configured, which the EdgeMarc is able to translate. Because the phone includes the softswitch address in its SIP message, the EdgeMarc is able to forward the SIP message to the softswitch. The proxy mode is multi-homed, because it is possible for phones connected to a single EdgeMarc to be served by different softswitches. In addition to performing translation, the EdgeMarc is able to forward the signaling to the correct softswitch for each phone.

In transparent mode, the phone is configured with the IP address of the softswitch, and Layer 3 packets are sent all the way from the phone to the softswitch without being terminated. The EdgeMarc picks up the signals as they are forwarded through the appliance and translates, but does not terminate them. Because the EdgeMarc hides the local topology from the WAN, the softswitch does not know the private IP addresses of the phones attached to the appliance; however, the phones know the identity of the softswitch. For this reason, it is also possible for different phones connected to the EdgeMarc to be served by different softswitches.

Survivability Mode

Survivability capabilities are triggered if the EdgeMarc loses its connection to the WAN or if the network softswitch is not accessible. When this occurs, the EdgeMarc uses the information that was obtained from the terminated or forwarded phone signals to direct local calls between phones attached to the appliance.

For example, assume that phones A and B are connected to the EdgeMarc and that A attempts to call B. The Survivability process works as follows:

1. Phone A sends a request to the EdgeMarc.
2. The EdgeMarc attempts to forward the call to the softswitch IP address that is configured in the appliance. The number of attempts is dictated by the Survivability settings.

3. If the attempt to contact the softswitch fails, the EdgeMarc takes over and performs the softswitch function, connecting phone A to phone B.
   
   

   
   

   Note

   The EdgeMarc always attempts to connect to the softswitch configured in the appliance. In transparent mode, this could be different from the softswitch configured in the phones. If the softswitches are different, the EdgeMarc is unable to perform the Survivability function.