What is FCIP, and what is it used for?
Fibre Channel over IP, or FCIP, is a tunnelling protocol used to connect Fibre Channel (FC) switches over an IP network, enabling interconnection of remote locations. From the fabric view, an FCIP link is an inter-switch link (ISL) that transports FC control and data frames between switches.
FCIP routers link SANs to enable data to traverse fabrics without the need to merge fabrics. FCIP as an ISL between Fibre Channel SANs makes sense in situations such as:
- Where two sites are connected by existing IP-based networks but not dark fibre.
- Where IP networking is preferred because of cost or the distance exceeds the FC limit of 500 kilometres.
- Where the duration or lead time of the requirement does not enable dark fibre to be installed.
FCIP ISLs have inherent performance, reliability, data integrity and manageability limitations compared with native FC ISLs. Reliability measured in percentage of uptime is on average higher for SAN fabrics than for IP networks. Network delays and packet loss may create bottlenecks in IP networks. FCIP troubleshooting and performance analysis requires evaluating the whole data path from FC fabric, IP LAN and WAN networks, which can make it more complex to manage than other extension options.
Protocol conversion from FC to FCIP can impact the performance that is achieved, unless the IP LAN and WAN are optimally configured, and large FC frames are likely to fragment into two Ethernet packets. The default maximum transfer unit (MTU) size for Ethernet is 1,500 bytes, and the maximum Fibre Channel frame size is 2,172 byes, including FC headers. So, a review of the IP network’s support of jumbo frames is important if sustained gigabit throughput is required. To determine the optimum MTU size for the network, you should review IP WAN header overheads for network resources such as the VPN and MPLS.
FCIP is typically deployed for long-haul applications that are not business-critical and do not need especially high performance.
Applications for FCIP include:
- Remote data asynchronous replication to a secondary site.
- Centralised SAN backup and archiving, although tape writes can fail if packets are dropped.
- Data migration between sites, as part of a data centre migration or consolidation project.
This was first published in August 2011