Cisco Tips & Tricks

November 17, 2009

L2VPN/VPLS-Martini and Kompella

Filed under: bgp, ccie, cisco, MPLS — ciscotips @ 2:24 am

Both Martini-draft and Kompella-draft addressed setting up of a Pseudowire emulation over MPLS in order to offer L2VPN services. These drafts were initial efforts to standardise L2VPN services.

Martini draft was named after a former Cisco employee Luca Martini. Martini draft uses LDP as signalling to setup L2VPN over MPLS backbone. The tradeoff of this draft was auto-discovery.

Kompella draft on the other hand uses BGP for both signalling and auto-discovery to establish fully-meshed pseudo wires (multipoint). Kompella-draft is named after author Keerti Kompella (Juniper Employee).

draft-martini and draft-kompella terms are used as labels for the two different L2VPN services methodologies (LDP Vs BGP for signaling). The actual drafts do not exist in IETF.

In dealing with multipoint-fully meshed topologies in edge routers, draft-martini suffered auto-discovery, to overcome aut0-discovery, it suffered configuration overhead. draft-Kompella claimed to be better scalable because of suto-discovery but with complex signalling whereas draft-martini leverages simplicity.

Martini draft was standardized under RFC 4096 . however it has since been superseded by the Pseudowire Emulation Edge to Edge (PWE3) Working Group specifications described in RFC 4447 and related documents. On the other hand draft-kompella is obsolete and was not standardized..

 RFC 4664 – Framework for Layer 2 Virtual Private Networks (L2VPN), it describes the framework for L2VPNs (VPWS, VPLS and IPLS). This framework is intended to aid in standardizing protocols and mechanisms to support interoperable L2VPNs. Requirements for L2VPNs can be found in RFC 4665 – Service Requirements for Layer 2 Provider-Provisioned Virtual Private Networks.

All this was consolidated, and the L2VPN Working Group produced two separate documents, RFC 4761 and RFC 4762, both offered VPLS but using different signaling protocols:

Kireeti Kompella and Yakov Rekhter published “Virtual Private LAN Service (VPLS) Using BGP for Auto-discovery and Signaling” RFC 4761 in January 2007.

Marc Lasserre and Vach Kompella published “Virtual Private LAN Service (VPLS) Using Label Distribution Protocol (LDP) Signaling” RFC 4762 in January 2007.

L2VPN services for many vendors uses RFC 4762 -Martini ( with LDP) as a standard for example Alcatel 7450’s uses RFC 4762 as the standard

November 8, 2009

CCIE notes for GLBP

Filed under: ccie, cisco, Technology and Software — ciscotips @ 5:47 am

Gateway load balancing protocol performs similar function to HSRP and VRRP. In both HSRP and VRRP,  group of routers participating in first hop-redundancy has one Active and can have multiple Client routers. At one single time, traffic is being passed through Active router, leaving client routers with unused bandwidth. Client routers will only become active once Active router in a group fails. We can create multiple groups and create different active routers but it results in extra administrative burden.

GLBP on the other hand can provide load balancing over multiple routers (gateways) using a Single Virtual IP and multiple Virtual mac-addresses. The bandwidth/traffic load is shared between multiple routers participating in the group rather than being handled by a single active router.

Following are the important points conceptually for GLBP.

  1. GLBP uses single Virtual IP and multiple mac-addresses to provide first-hop Gateway redundancy.
  2. In GLBP, there can be four routers/gateways in a group
  3. Hello messages are used to communicate with in the group destined to 224.0.0.102, udp port 3222 and they will be sent every 3 secs by default.
  4. initially group members will elect one AVG ( Active Virtual Gateway) and other routers will act as backup AVG’s incase the active AVG fails
  5. AVG will assign Virtual mac-addresses to other routers, they are known as AVF’s ( Active Virtual Forwarders)
  6. Each AVF assumes responsibility for  forwarding  packets sent to Virtual Mac’s assigned by AVG.
  7. AVG is responsible for answering ARP requests for Virtual IP’s

Configuring GLBP

R2(config-if)#glbp 1 load-balancing ?
  host-dependent  Load balance equally, source MAC determines forwarder choice
  round-robin     Load balance equally using each forwarder in turn
  weighted        Load balance in proportion to forwarder weighting

There are three different types of Load balancing algorithms in GLBP.

Host-Dependent

  1. The Mac-address of the host is used to determine which AVF’s  mac is the host directed towards.
  2. A given host is guaranteed to use the same Virtual Mac as long as number of VF’s in the GLBP group are constant
  3. Host dependant GLBP is not recommended in situation where there are small number of hosts, for example, less than 20

Weighted

  1. GLBP places a weight on each device to calculate the amount of load sharing that will occur through MAC assignment
  2. Each GLBP router in a group will advertise its weight and AVG will act based on that value
  3. For example  we have two routers, Router A and Router B. If router A has double the bandwidth capacity then router B. Router A will be configured with the double weighting value of router B

Round-Robin

  1. With Round-robin VF mac-address is used sequentially in ARP replies for the virtual IP
  2. This is the default type of GLBP algorithm
  3. It is suitable for any number of hosts.

Steps for  configuring GLBP

  1. enable GLBP with glbp 1 load-balancing
  2. glbp 1 priority ( Higher is better, default is 100)
  3. glbp 1 ip x.x.x.x
  4. glbp 1 preempt < To enable preempt, by default its disabled>
  5. glbp 1 authentication  ( Enabling authentication with in a group)

Verification

Show glbp

Create a free website or blog at WordPress.com.