IP Switching 國立中正大學 資訊工程研究所 黃仁竑 副教授

中正資工 / 黃仁竑 2 IP Switching o Problem with classical IP over ATM ê IP over ATM preserves ATM protocol stack as well as TCP/IP protocol stack ê IP routing protocol running at IP layer ê ATM signalling running at ATM control plan o Do we have other choices ê Discard ATM signalling/routing ë Ipsilon IP switching, Cisco Tag switching ê Incorporating IP routing with ATM routing ë Ascend IP Navigator, IBM ARIS o Issues of IP switching ê Switching and routing ê Flow classification ê QoS support ê Multicast support

中正資工 / 黃仁竑 3 Ipsilon IP Switching o Run IP over ATM hardware IP AAL ATM MAC ATM Switch IP ATM Switch Ipsilon

中正資工 / 黃仁竑 4 Protocols o IFMP ê When downstream node identifies a flow, send flow identifier to the upstream node which indicates which VPI/VCI should be used for the flow. o GSMP ê For the IP switch controller to communicate with ATM switch

中正資工 / 黃仁竑 5 Flow Classification o What is a flow ê A sequence of IP packets that belong to the same IP service ê “extended IP conversation” o Flow characterization ê Same source-destination pair ê Same protocol type (TCP/UDP) ê Same type of service (port number) ê Flow label in IPv6 o Cut-through ê Long duration flows can be optimized by cut-through switching in the ATM hardware. ê The rest of the traffic continues to receive the default treatment hop-by- hop store-and-forward routing. o Homogeneous Ipsilon IP switches ê Recognize flow locally, but if all with the same criteria, an end-to-end ATM switching path will be built

中正資工 / 黃仁竑 6 Flow Type o A host-pair flow type (flow type 2) ê For traffic flowing between the same source and destination IP addresses. o A port-pair flow type (flow type 1) ê For traffic flowing between the same source and destination TCP/UDP ports on the same source and destination IP addresses. ê The port-pair flow type allows quality of service differentiation among flows between the same pair of hosts and also supports simple flow-based firewall security features.

中正資工 / 黃仁竑 7 IFMP Note: When flow is identified and VC is set up, no LL/SNAP encapsulation is required.

中正資工 / 黃仁竑 8 GSMP o Five types of message ê Configuration: discover the capabilities of the ATM switch ê Connection management: establish/remove connections across switch ê Port management: reset, bring up, take down, and loopback switch ports ê Events: asynchronously alter the control significant events ê Statistics

中正資工 / 黃仁竑 9 IP Switching Operations IP Switch Controller ATM Switch Upstream Node Upstream Node Downstream Node Downstream Node IP Switch  Connectionless packets are forwarded over default ATM VCs and IP switch controller makes a flow classification on each packet 

中正資工 / 黃仁竑 10 IP Switching Operations IP Switch Controller ATM Switch Upstream Node Upstream Node Downstream Node Downstream Node IP Switch    IP switch controller sends a message to the up-stream node to use a new VC for a selected flow.  Traffic for the selected flow begins to flow on the new VC

中正資工 / 黃仁竑 11 IP Switching Operations IP Switch Controller ATM Switch Upstream Node Upstream Node Downstream Node Downstream Node IP Switch    Downstream node will also request a new VC for the flow  IP switch begins to send traffic on that flow to the downstream node on the new VC

中正資工 / 黃仁竑 12 IP Switching Operations IP Switch Controller ATM Switch Upstream Node Upstream Node Downstream Node Downstream Node IP Switch  Incoming labeled flow switched through to outgoing labeled flow where “cut-through” operation completed for flow-oriented traffic.

中正資工 / 黃仁竑 13 Miscellaneous Issues o Multicast ê Support IP multicasting without any modification to IGMP ê Can utilize Switch’s multicast functionality ê Identify multicast flow based on source-based (point-to-multipoint) tree o QOS ê Basically, lack of QOS since no “real” VC is set up. May cooperate with RSVP in the future ê It’s up to ATM switch o Robustness ê Each IFMP redirection is associated with a timer ê New IFMP redirection must be sent before timeout if the flow continues

中正資工 / 黃仁竑 14 Tag Switching o A new technique, developed by Cisco, for high-performance packet forwarding that assigns "tags" to multiprotocol frames for transport across packet or cell-based networks. o Based on the concept of "label swapping," in which units of data carry a short, fixed length label that tells switching nodes how to process the data.

中正資工 / 黃仁竑 15 Tag Switching Internetwork

中正資工 / 黃仁竑 17 Tag Switching Internetwork o Tag edge routers: ê located at the boundaries of an Internet, tag edge routers perform value-added network layer services and apply tags to packets. o Tag switches: ê switch tagged packets or cells based on the tags. Tag switches may also support full Layer 3 routing or Layer 2 switching, in addition to tag switching. o Tag distribution protocol (TDP): ê in conjunction with standard network layer routing protocols, TDP is used to distribute tag information between devices in a tag switched Internet.

中正資工 / 黃仁竑 18 Tag Switching Operations o Step 1: Tag edge routers and tag switches use standard routing protocols to identify routes through the network. ê Fully interoperable with non-tag switching routers. o Step 2: Tag routers and switches use the tables generated by the standard routing protocols to assign and distribute tag information via the TDP. o Step 3: Tag routers receive the TDP information and build a forwarding database which makes use of the tags. o Step 4: When a tag edge router receives a packet for forwarding across the tag network, it analyzes the network layer header selects a route for the packet from its routing tables, applies a tag, and forwards the packet to the next hop tag switch.

中正資工 / 黃仁竑 19 Tag Switching Operations o Step 5: The tag switch receives the tagged packet and switches the packet based solely on the tag, without re-analyzing the network layer header. o Step 6: The packet reaches the tag edge router at the egress point of the network, where the tag is stripped off and the packet delivered.

中正資工 / 黃仁竑 20 Tag Switch Components o Forwarding component ê Uses the tag information carried by packets and the tag forwarding information maintained by a tag switch to perform packet forwarding o Control component ê Creating tag bindings, and then distributing the tag binding information among tag switches

中正資工 / 黃仁竑 21 Forwarding component o Tag Information Base (TIB) - each entry consists of : ë Incoming tag ë One or more sub-entries (outgoing tag, outgoing interface, outgoing MAC address) o Forwarding algorithm ë Based on the exact match algorithm ë Independent of the tag’s forwarding granularity ë Could be implemented with any MAC/link layer technology ë Network layer independent o Carrying tag information ë As part of the network layer header (IPv6) ë As part of the MAC header (VCI/VPI in ATM) ë Via a “shim” between the MAC and the network layer header

中正資工 / 黃仁竑 22 Control component o Organized as a collection of modules, each module is designed to support a particular routing function : ê Destination-based routing ê Hierarchy of routing knowledge ê Resource reservation ê Explicit routes ê Multicast o New modules could be added to support new routing functions without impacting the forwarding component

中正資工 / 黃仁竑 23 Destination-Based Routing Module o Forwarding decision is based on the destination address carried in a packet and the information stored in the Forwarding Information Base (FIB) o A tag switch constructs its FIB by using the information receives from routing protocols (e.g., OSPF, BGP) o Three methods for tag allocation and Tag Information Base (TIB) management ê downstream tag allocation ê downstream tag allocation on demand ê upstream tag allocation

中正資工 / 黃仁竑 24 Downstream Tag Allocation A For each route in its FIB the switch allocates a tag, creates an entry in its Tag Information Base (TIB) B Advertises binding between the incoming tag and the route to all of the adjacent switches : by either piggybacking the binding on top of the existing routing protocol, or by using a separate Tag Distribution Protocol (TDP) C When a switch receives tag binding information for a route, if the information was received from the next hop for that route, the switch places the tag into the outgoing tag of the TIB entry associated with the route

中正資工 / 黃仁竑 25 Downstream Tag Allocation on Demand A For each route in its FIB, the switch request (via TDP) the next hop for a tag binding for that route B When the next hop receives the request, it ê allocates a tag ê creates an entry in its TIB with the incoming tag set to the allocated tag ê returns the binding to the requester C When the requester receives the tag binding information for a route from the next hop for that route, the requester places the tag into the outgoing tag of the TIB entry associated with the route

中正資工 / 黃仁竑 26 Upstream Tag Allocation A If a tag switch has one or more point-to-point interface, then for each route in its FIB whose next hop is reachable via one of these interfaces ê The switch allocates a tag ê Creates an entry in its TIB with the outgoing tag set to the allocated tag ê Advertises to the next hop (via TDP) the binding B When the next hop receives the tag binding information, the switch places the tag into the incoming tag of the TIB entry associated with the route

中正資工 / 黃仁竑 27 Hierarchy of Routing Knowledge Module o Allows the de-coupling of interior and exterior routing ê Between domains use tags with exterior routes (BGP tag) ê Within a domain use tags associated with interior routes to BGP border routers of the domain (IGP tag + BGP tag) ê Tag (label) stack o Reduces the routing load on non-border switches o Shortens routing convergence time

中正資工 / 黃仁竑 28 Explicit Routes Module o Overrides the hop-by-hop destination-based routing paths o Requires the ability to install tag bindings that are independent from the tags installed via the destination-based routing protocol ê May be coupled with resource reservations o Possible applications : ê Allows finer control over traffic distribution over multiple paths ê Support forwarding in QoS-based routing