iscsi failover and a "ping node"?

So, I've been looking through some of the literature I was sent about iscsi failover for open-e, and some of it mentions something called a "ping-node."

Is this required? How is a ping node set up? Is this a common feature of high-end switches?

One important thing is that this might lead to a "single point of failure." Is this true? If a ping node is required for auto failover, can a ping-node have a hot spare of some sort, so there wouldn't be a single point of failure?

We have customers who are asking these sorts of questions, and we don't know how to answer them. Can you help?

Dear Robotbeat,

Its a good chance to explain what and how configure the ping node.

First, Ping node can be a server, router or any device that can be pinged.

And here how it's working:

Lets say you have 2 servers (A&B) you connect them thru connection 1, then you have a second connection from each server to the ping node.

You set your replication task and your virtual IP address. Then, start the Autofailover. Now, lets test those scenarios:

1- If the connection 1 failed. Server A is going to check with the Ping node if server B is up or not. Or if its just the connection get dropped.

2- If the connection failed between server A and the ping node. Server B is going to check thru connection 1 and see if the server A is up.

3- If the connection between B and ping node failed, server A will check if B is up from connection 1.

You can see here that the Ping node is the Heart Beat of the process. And its not a single point of failure. Actually, it provides reliability for the process.

Let me know if its not clear yet.

Regards,
SJ