MikroTik RouterOS Training Inter-Networking Overview + BGP + BGP Basics (iBGP, eBGP) * Distribution, filtering and BGP attributes + MPLS * Introduction to MPLS + LDP + L2 and L3 VPN's * Traffic EngineeringLab Setup + Divide in groups by four + Make network setup as illustrated in next slide + R1 and R2 routes connect to AP with SSID AS100 in 2.4Ghz band + Each router has local network [Link].0/24 where: + X-group number + Y-Routers number X- group number Lab Setup AP SSID“A100 band=2.4Ghz ast00Border Gateway Protocol (BGP) Autonomous system + Set of routers under a single administrative control + Routing exchange: + Routers within AS use common IGP + Routers between ASs use EGP * Has its own number (ASN) + Supports 16-bit value and 32-bit value + Numbers 64 512 — 65 534 reserved for private useBGP Basics + Stands for Border Gateway Protocol + Designed as Inter-AS routing protocol + Network topology is not exchanged, only reachability information. Only protocol that can handle Internet's size networks Uses path vector algorithm Path Vector Implementation + Treats whole AS as a single point in the path * Prefix is advertised with the list of ASs along the path called AS path + Hides network topology within an AS + Cannot provide loopfree routing within an ASPath Vector Implementation ‘Add ASI00 [Link]'24 otepeh asi a 8200 Reject. AS 100 f areadyin 4 thepath } nda.asz0o J tothe path tothe path BGP Capabilities BGP Speaker advertises supported capability codes If received capability is not supported, remote peer sends back notification BGP speaker attempts to peer without unsupported capability Some of RouterOS advertised capabilities: + Route refresh + Multi-protocol extension + 4-byte AS supportBGP Transport Operates by exchanging NLRI (network layer reachability information). NLRI includes a set of BGP attributes and one or more prefixes with which those attributes are associated Uses TCP as the transport protocol (port 179) Initial full routing table exchange between peers Incremental updates after initial exchange (maintains routing table version) Packet format Packet contains four main fields: + Marker (128bits) — used for authentication + Length (16bits) + Type (8bits) - BGP message type + Message bodyBGP message types + Four message types: + Open — First message sent after TCP connection establishment, contains capability list. Confirmed by keepalive. + Keepalive - does not contain data, sent to keep hold timer from expiring + Update — actual route updates. Contains: ~ NLRI ~ Path attributes * Notification — sent when error condition occurs, contains error code and sub-code BGP session and updates Open with ASN4 capability ‘AS100 “Notification unsupported cap. Koopalive Route Refresh message UpdateNetworks Indicates what networks BGP should originate from the router. By default network is advertised only if corresponding route is present in routing table Synchronization can be turned off if: + Your AS does not provide transit service + Allthe transit routers run BGP Disabling sync allows BGP to converge faster. Sync can be dangerous if routes are flapping a lot. Configurable from /routing bgp network Enable BGP Jeoating bgp instance /routing bap peer If router is not specified, its automatically set to least IP address on the router. Verity BGP connectivity. Any state other than established indicates that routers can not become neighbors (use print status for more details)Stub network Scenarios + Single homed Private ASN is used (764511) ~ ISP originates only default route Actually no need for BGP ~ Upstream ISP advertises networks ~ Stub network has the same policy as ISP 1s ‘stub net ee | ss00 Private AS Removal Global net g 7ASGS500 wn anise * Private AS cannot be \ ~~ leaked to public Ke Bases * Available for eBGP ‘zoo UN massa neighbors p. + Announce only aggregate route —_,., 86? * Use following command oa Boe oeBGP Lab | * Create BGP network setup as illustrated in next slide: + BGP peer from R1 and R2 to AP. + BGP peer from R2 to R4 * BGP peer from R1 to R3 + Advertise your local network + Private ASN should be removed * Originate default route to private AS routers BGP Lab | X~ group number [AP SSID“AS100 band=2.4Ghz ==" BGP peor taaenxto2Stub network Scenarios + Multihomed Private ASN is used ~ Can be used: + As mainfbackup link + Load sharing Upstream ISP advertises networks Stub network has the same policy as ISP 1p 2 sane Basen tone me Non-stub Scenarios Need to obtain AS number from ISP or RIR Address range from Regional Internet Registry Routing policy independent from ISPs Can be used: As main/backup link ~ Load sharing a “e ~ More advanced AS100 AS200 routing policies ‘ p ‘88300 Global netBGP and connection tracking + Connection tracking is unable to keep valid track of connections with multi-homed BGP. + Packets related to one connection can travel through different paths ~ Do not drop invalid connections in firewall * Con-track should be turned off for better performance BGP Lab II + Add R3 to the same AS as R1 + Add R4 to the same AS as R2 + Make BGP peer between R4 and R3 + Set up OSPF between routers in the same AS * Set OSPF to distribute connected routes * Announce both local networks from ASBGP Lab II [AP SSID“AS100 band=2.4Ghz ==" BGP peor aaenxt.02 As100 BGP Lab II [adninéRt] /ip route> print Flage: X - disabled, A'~ active, D - dynamic, C- connect, §- static, r- rip, b- bgp, 0 - opf, = - ane, 1B - blackhole, U - unreachable, P ~ prohibit [Link]/0 [Link] 20 [Link]/24 [Link] RAP [Link]/30 192.1682. RIS i92/168.1.6/30 392 .168.1.2 192/168.11.0/24 [Link] local i92/168.11.0/24 [Link] i92/168 12/0/24 92/168. 10 1921168,12.0/24 [Link] i92-168.13.0/24 [Link] 92/168 113.0/24 392 .368.1.2 1921168 ,14.0/24 392/168 .1.10 192/168,14.0/24 [Link] pbebebyRPRREBGP Lab II /4p route? print Flage: X - disabled, A ~ active, D - dynamic, c= connect, § - static, r- rip, b- byp, o- ospf, m- me, B - blackhole, U = unreachable, P ~ prohibit [Link]/0 392.168 200 [Link]/24 92.168 no [Link]/30 192,168.22 -R3 AL ° i92'68.1.6/30 © [Link] RIE ° 1921168.11.0/24 192,168 no i92/168.11.0/24 392.168 200 192/168 .12/0/24 92 168 20 1921168.13.0/24 [Link] local ° 492/168 .13.0/24 92-268 200 i92/168 1410/24 92/168 20 BeRbePRERE ‘ouder Rt does nt redsriue 12004 and 404 back to RO Interior and Exterior BGP + BGP — peering between routers inside an AS + eBGP — peering between routers from different ASseBGP + Almost always formed between directly connected peers (AS edge routers). + Multi-hop configuration is required if peers are not directly connected + Adds AS to advertised prefix's path + By default Next-hop is changed to self eBGP Multihop example ewsaf e-as-x multihopayes Configuration requires static routes or enabled IGP so that the neighbors ccan reach each other. Selting @BGP to Loopback addresses can protect BGP from DOS attacksiBGP Next-hop is not changed by default: + Uses IGP (RIP,OSPF;static) to ensure network reachability within an AS Attributes learned from iBGP are not changed to impact the path selection to reach outside network AS path is not manipulated Provides ways to control exit point from an AS. Received external route from iBGP peer is not propagated to other iBGP peers: + Requires full mesh between iBGP peers. 8 Db 192.168 [Link] 200 9 an 192/160 114.0/24 20.20.0112 20 8 Db dst-addross=192.168,14.0/24 gateway=[Link] gatoway-stat target-scape bbgp-as-path="100,112" bop-oign=igp received{rom=pear!Loopback + Eliminates dependency from physical interface to make TCP connection. + Mostly used between iBGP peers + In RouterOS empty bridge can be used as loopback ‘gore a - [interface bridge (sp address a3 3 ‘freuting bgp peer BGP Lab III + Improve your setup by using loopback addresses between iBGP peers 10.255.x.y/32, where Xx — group number y — router's number + Add loopback address to OSPF networks * Set loopback address as ospf and bgp router-idRoute Distribution + IGP (Static, OSPF, RIP, connected) routes can be redistributed {routing bap instance * Prefix origin is “incomplete” + Risk of advertising all IGP routes + Always use routing filters to avoid unwanted route advertisements Distribution Example [sp route sad dst ‘/routing bgp instance + Packets will be ceased unless more specific route is present * Good way to advertise supernetRouting Filters * Main tool to control and modify routing information Organized in chains similar to firewall Specify in BGP peer’s configuration which chains to use or BGP instance out filter Prefix passes instance chain, then moves to peer's chain /rosting bop peor bap-in ou out [routing f£lter osc ee ‘/routing bap peer set Jrouting £ target-scope Dbgp-commurPrefix filtering ies p f contig on R3 /routing bap peer se /routing filter AS Path filtering Can be configured to allow updates only to/from certain AS Supports regular expressions +“ sany single character start of the as-path end of the as-path matches comma, space, start and end of as-path 1 contig on 12 /routing filer ads nailBGP Soft Reconfiguration + When action=discard is used, routes are not updated after filter change. * Solution + Use action=reject to keep routes in the memory + Dynamic (Peer must support refresh capability) ~ Peer refreshes the routes afer the changes are done. ~ No additional memory is used ~ Itis not done automatically - need to run “refresh” command BGP Lab IV * Set up routing filters in the way that: + R1 does not receive 192. 168.x2.0/24 prefix via AP + R2 does not receive 192.168.x1.0/24 prefix via AP + R3 does not receive 192.168.x4.0/24 prefix from R4 + R4 does not receive 192.168.x3.0/24 prefix from R3BGP Lab IV Lets look at R3. If set up properly traceroutes to network x2 should go over R4 and traceroutes to x4 should go over AP BGP decision algorithm + BGP uses single best path to reach the destination + BGP always propagates the best path to the neighbors + Different prefix attributes are used to determine best path, like weight, next-hop, as-path, local- pref etc. * Setting peer to loopback address can force BGP to install ECMP route (for load balancing)