The three most popular classes of service for SANs in a Fibre Channel fabric are 1, 2 and 3. Class 1 is a virtual direct connection through the SAN between the transmitting and receiving devices, class 2 uses packets which are acknowledged transmission (either after each packet or every N packets) and can follow multiple paths through the switch fabric. Class 3 also uses multiple paths but doesn't use the Fibre Channel acknowledgment. Detection and retransmission of corrupted information are handled by a higher-level protocol like IP rather than through Fibre Channel. All three classes are popular because they all serve different needs.
If you're going to do a lot of multicasting, then Class 3 Fibre Channel service is for you. It's a connectionless service using frame switching, like Class 2. However unlike Class 2, Class 3 is designed around a datagram model and Fibre Channel doesn't check for missing or corrupted frames. Flow control and data integrity are handled by a higher-level protocol such as TCP (Transmission Control Protocol). However there is link-level flow control between the fabric and each connected device.
In multicasting, the transmitting device sends the same information to multiple receiving devices. Since the receiving devices do not send back acknowledgments, rejections or other messages, this service cuts the load on the transmitting device considerably. If the higher-level protocol on the receiving device
The disadvantage of Class 3 service is, of course, the lack of acknowledgments. This is compounded by the fact that in Class 3 service the fabric may occasionally discard some frames when traffic is heavy. The higher-level protocol at the receiving device can ask for a retransmission of the sequence, but that degrades device and fabric throughput.
- For earlier SAN/NAS Technical Tips discussing Fibre Channel Class 1 and 2, please see: http://searchstorage.techtarget.com/searchStorage_Tips_Category_Page/0,1797,476,00.html
- Brocade Communications Systems has a white paper discussing classes of service and their effect on data transport on its web site at: www.brocade.com/SAN/white_papers/data_transport.html
About the author: Rick Cook has been writing about mass storage since the days when the term meant an 80K floppy disk. The computers he learned on used ferrite cores and magnetic drums. For the last twenty years he has been a freelance writer specializing in storage and other computer issues.
This was first published in December 2000