AllPath Bridging Update Jun Tanaka Fujitsu Labs Ld
All-Path Bridging Update Jun Tanaka (Fujitsu Labs. Ld. ) Guillermo Ibanez (UAH) Vinod Kumar (Tejas Networks) IEEE Plenary meeting San Francisco 17 -21 July
All-Path : Basic Idea x Blocking x MAC=a Flooding MAC=b • Modified learning, filtering and forwarding • Addressed to small and medium size networks • Objectives: overcome RSTP limitations and obtain good latency paths without configuration
Scope of All-Path Both support, loop free, shortest path Manageability SPB, ECMP TRILL Enterprise, Campus, Small datacenter etc. LAN ALL-PATH l Simple l Less operation l Natural load balance Large area, provider network Large datacenter etc. MAN/WAN Scalability 2021/12/13 2
Issue 1. Path Recovery • Mechanism: When unknown unicast frame arriving at bridge with failed link, path fail message is generated per MAC entry towards source bridge, that generates corresponding ARP to re-establish tree. • Question: If 10 K MAC entries are existed in FDB, 10 K path fail frames should be generated, is it feasible processing for local CPU, especially in high-speed link (ex. 10 GE)? • It can be applicable for small or medium networks if we are targeting the enterprise, campus and small datacenter etc. • Moreover, a solution is proposed that minimize the frame generation. 2021/12/13 3
Path recovery (cont’d) a a sh ” “b flu 3 a 1 MAC=a flush “b” 2 1 SW 5 2 flush message is terminated because “b” is not binded to port 1 a 1 b 2 SW 1 1 SW 3 b 2 SW 2 Delete entry “b” from FDB and re-sends the flush message to SW 1. a 1 flush “b” b 2 SW 4 a 1 3 b 2 SW 6 MAC=b May includes two or more…ex. 100 s of MAC addresses to be flushed as a list. l When link failure is detected, MAC flush lists (187 MAC / 1500 B frame) are flooded. 54 frames by a day for 10 K MAC entry. l Avoid unnecessary flooding, MAC list is deleted to shorten. l Issues: How to prevent flush frame loss. May require CPU processing power. l Experience: 15 ms to flush 10 K MACs in a node (1 GHz MIPS Core) 2021/12/13 4
Issue 2. Server Edge • Question: If a server has two or more NICs, how to find which port is first? • vswitch: only vswitch to support All-Path • VEB: both VEB and vswitch to support All-Path • VEPA: only external switch to support All-Path Vswitch NIC Vswitch VEB NIC VEPA NIC Ext. switch 2021/12/13 5
Issue 3. Load Balance • Load balance is available in natural way because high load link tend not to be selected with queuing delay. • Pros: zero-configuration load balance, no extra processing effort • Cons: you cannot control load balance like SPB/ECMP 2021/12/13 6
Report on Proposal for AVB TG • May 12, Thu, morning session @ AVB • Dr. Ibanez presented the materials as used in IW session (Singapore and Santa Fe) • Questions and comments – – Any other metric than latency e. g. bandwidth? Path recovery time comparing with RSTP? Any broadcast storm occurred when link failed? What’s the status in IW session, any PAR created? • AVB status – They try to solve by their own way, using SRP. – Not only latency but also bandwidth can be used as metric – Also redundant path can be calculated 2021/12/13 7
- Slides: 8