HAIR: Hierarchical Architecture for Internet Routing Anja Feldmann TU-Berlin / Deutsche Telekom Laboratories Randy Bush, Luca Cittadini, Olaf Maennel, Wolfgang Mühlbauer

Routing scalability: Problems r IP addresses usage m Locator within the Internet m Identifier for applications r Routing table size growth m Multi-homing m Traffic engineering m Prefix disaggregation

Routing scalability: Problems r Churn: High update rates m Due to mobility m Due to global visibility m Due to „overuse“ of policy m...

Routing scalability: Current workarounds static large RT high upd rate high upd rate dynamic Scalability issues expensive TCAM high workload to maintain RT high workload to maintain RT control plane data plane Consequences limited TE limited TE limited mobility limited mobility Problems massive filtering massive filtering dampening Workarounds static dynamic

Approach r Key ideas m Separation of locator/identifier function of IP address => separation of routing and location mapping TU-Berlin

Approach r Key ideas m Separation of locator/identifier function of IP address => separation of routing and location mapping m Hierarchy for routing and location mapping

Approach r Key ideas m Separation of locator/identifier function of IP address => separation of routing and location mapping m Hierarchy for routing and location mapping r Two components m Routing system based on locator m Mapping system to map an identifier to a locator

Hierarchical routing r Network is organized in multiple levels r Levels are separated by separators r Routers only know the details about their level Separator

Hierarchical routing: Internet r Where do we have small separators? r Internet structure m Core  Set of interconnected autonomous systems (ASs)  Tier-1, tier-2 ASs, …  Transit ASs

r AS core m ~5000 ASs r AS edge m ~30000 AS Stub AS Access Provider Core Enterprise Network ISP1 ISP2 ISP3 Transit AS 1 Transit AS 2

r AS core m ~5000 ASs r AS edge m ~30000 AS Stub AS Access Provider Core Enterprise Network ISP1 ISP2 ISP3 Transit AS 1 Transit AS 2 Potential large separator Potential small separator

Hierarchical routing: Internet r Where do we have small separators? r Internet structure m Core  Set of interconnected autonomous systems (ASs)  Tier-1, tier-2 ASs, …  Transit ASs m Intermediate  Stub ASs, e.g., metropolitan area networks  Enterprise networks  Content distribution networks m Edge  Local area networks

Hierarchical routing: Internet r Separator size m Core -> Intermediate  Stub ASs, e.g., metropolitan area networks: < 10 links  Enterprise networks: < 10 links  Content distribution networks: < 1000 links m Intermediate -> Edge  Local area networks: < 10 links r Terminology m Core /WAN m Intermediate / MAN m Edge / LAN m Separator / Attachment point (AP)

Hierarchical network r Example: Three levels of hierarchy m Routing via intermediate points – the separators => specify attachment points m WAN APs: WAP  Provider access links m MAN APs: MAP  Firewalls

Sending a packet r Routing via intermediate access points m Mapping service: resolve identifier to locator m 3 locator parts: WAP|MAP|ID

Routing scalability r Core m Routing based on WAPs m Stable business relationships m Almost no churn m Aggregatable addresses m Common routing protocol (e.g., BGP) r Intermediate (smaller ISPs/enterprises) m Routing based on MAPs m Separate addresses and routing m Local changes  local impact r Edge (e.g., Ethernet LAN) m Standard L2 switching

Mapping system r Design requirements m Scales with number of hosts m Fast response times m Easy to update r Approach m Clients are responsible m Hierarchical design  Global DHT or DNS like system –For each identifier: pointer to MMS –WANs contribute resources  MAN mapping service (MMS) –Stores locators for attached nodes –Provided by MAN(s)

Mapping identifiers to locators r Steps m Client queries  Global DHT  MMS r To avoid lookups m Use caching m Include source locators in packet m … r Global DHT/MMS m Can store multiple alternatives r Failure recovery m Via multiple alternatives

Discussion (1) r Scalability m Hierarchical routing AND mapping system m Updates are localized => low update rates m No manual configuration r Mobility: local visibility of changes m Intra-MAN mobility: frequent  Updates restricted to MMS m Inter-MAN mobility: less frequent  Update global DHT (fast)  Move locators to new MMS

Discussion (2) r Multihoming m Inherent support: APs exposed to routing system r Multipath m Use multiple locators in parallel r Inbound traffic engineering m Per-host basis m MANs/MMS have control r Migration path m To support legacy hosts

Migration via NATs/Firewalls: Sending r Firewalls/NAT act as MAPs r Legacy packet arrives from LAN m Treat dst address as dst ID m Resolves locator for ID m Add source locator to packet header m Encapsulate original packet and sends it

B Migration: Receiving r WAP strips encapsulation r MAP/NAT strips the second layer m May get the mapping for the source locator r Packet is routed onward To: WAP To: MAP Loc(A) From: A To: B To: MAP Loc(A) From: A To: B A => Loc(A) From: A To: B

What’s different here r Routing hierarchy based on structure of the Internet m Smaller table sizes m Lower update rates r Mapping service is hierarchical m With local control and responsibility r Hosts are responsible for obtaining mapping r Incremental deployment possible

Summary r Main goals m Scalability m Support for multi-homing, TE, mobility, etc. m Smooth migration, support for legacy hosts r Key ideas m Separation of locator/identifier function of IP address m Hierarchical routing and location mapping scheme r Two components m Routing system based on locator m Mapping system to map an identifier to a locater