RE: iFCP

[Joshua Tseng <jtseng@NishanSystems.com>]

Marjorie,

<snip..snip>
> 
> Please explain how you've arrived at this notion of tunneling 
> (FCoverIP)
> requiring n*(n-1) "edge connects" and what you mean by "edge 
> connects".

I think what John means is that there is a n*(n-1) configuration problem
in the number of FCIP tunnels that must be set up between n number
of SAN islands.  Since FCIP does not (yet) have a dynamic tunnel
configuration mechanism, each gateway must be manually configured with
n-1 tunnels to each of the other SAN islands, where n is the number of 
SAN islands (not necessarily gateways).  This is okay when there are
only two SAN islands that need to be connected, but the amount of manual
configuration increases exponentially as the number of SANs increases.

> 
> Implementation experience is a pair of "edge devices" 
> (FCoverIP switching
> devices) for a WAN link between remote SANs.  This is a proximity
> requirement only - ie, you have to have a physical device at 
> a location to
> connect to the local physical wiring(s).  So if there are 
> several SANs at a
> physical location, it still requires only one FCoverIP device 
> (with N FC
> ports) to handle the WAN connection.

Multiple SANs at the same physical location do not need FCIP to
interconnect them.  So yes, I agree we are talking about a proximity
requirement as far as the number of gateways.  But the issue is the
amount of configuration required to set up n*(n-1) tunnel endpoints,
not the actual number of gateways.

Josh

> 
> 
> Marjorie Krueger
> Networked Storage Architecture
> Hewlett-Packard Storage Organization
> tel: +1 916 785 2656
> fax: +1 916 785 0391
> email: marjorie_krueger@hp.com 
>