SAN/LAN/WAN convergence Day Four: Wrapping it all up

In the last part of our feature on network convergence, Ian Yates explains tiered networks and more.

Yesterday we saw how running different network protocols need not be a terrifying prospect.

Nortel's Ryan Perera agrees - at least when it comes to converging SANs onto LANs which he does not think will be a major issue when compared with getting SANs converged onto your WAN links. "SAN has very little significance at LAN levels," says Perera. "The real issue is about using wide area networks to support SAN applications. Because it makes sense to be able to maximise existing transport infrastructure to support storage, that's how things are evolving. Fibre channel is by and large the most prominent storage protocol," says Perera. "The question is which transport protocol should we use over the WAN? Should we use DWDM (dense wavelength division multiplexing), should we SDH (synchronous digital hierarchy) or should we use Ethernet?"

There is no one size fits all, so it's best to look at this on a tiered basis. Tier one would be a top end financial institution with very stringent requirements for connectivity. "They need private networks," says Perera. "They need exclusive, separate networks and they need scalability to cater for unpredictable growth. So therefore they end up using high range dense wavelength division multiplexing (DWDM) solutions. That market we think will continue to stick with a DWDM solution." Tier one businesses are also likely to have branch offices a considerable distance from the data centre, which would prohibit the use of expensive leased fibre links.

"Their DWDM solutions will get complimented by SDH because no-one is going to build an exclusive network between Sydney and Melbourne just for storage," says Perera. "They need to be able to use one of the existing long haul routes." So that's the tier one. But more interesting is tier two which includes businesses which can't afford to build their own private network using dark fibres and DWDM solutions because the the costs don't quite stack up. This tier really needs to make use of existing infrastructure.

"Now this is where storage over Ethernet really comes into play," says Perera. Tier two companies might also represent a branch office of a large tier one bank. "This segment definitely needs to rely on the existing infrastructure. The existing infrastructure that carries their LAN traffic, that carries their voice traffic and they want to be able to use that same infrastructure to carry SAN and the two mechanisms still at the early stages of their standardisation, is iSCSI and fibre channel. Where we have been focusing on is carrying fibre channel over Ethernet which we already do within our product portfolios."

As a provider of backbone infrastructure switching, Perera is somewhat agnostic about which protocol becomes the industry standard. "What we are focusing on is building carrier-grade Ethernet networks that have the traffic engineering capabilities, quality of service and the determinism so that it doesn't matter whether it is iSCSI or SDH over Ehternet or something else," says Perera. One thing seems certain, and that's the ever-encroaching reach of IP-based Ethernet to transport anything and everything to anywhere and everywhere. Storage will be just another passenger on the information super highway, albeit a passenger that almost always gets an upgrade to business class!


Read more on Data centre networking