Contents

Mobile backhaul Solution: IP-RAN Overview


This is a throwback blog when I was working as a Datacom engineer in a Huawei Technologies. This is a popular solution of IP based Radio Access Network (IP-RAN).


IP-RAN Overview & Technologies

  • As the mobile broadband technologies starts evolving to LTE and LTE advanced, industry/operators agreed that Radio Access Network (RAN) must be IP-based, hence, IP-bearer network aka IP-RAN network.
  • This MPLS VPN based IP-RAN also has requirements to meet bearer needs for fixed and mobile convergence to be able to carry multi-services, high reliability, effective O&M, clock synchronization, QoS.
  • IPRAN has MPLS/VPN with MP-BGP that can carry various services with secure tunnel.
  • IPRAN design considers a ring topology for aggregation and access devices to fulfill bandwidth requirements and service protection for high reliability.

Time Domain Multiplexing (TDM)

  • Before IP based transport layer, there used to be TDM technologies that carried all radio access network services like 2G, 3G or 4G.
  • SDH (Synchonous Digital Hierarchy) and MSTP (Multi-Service transmission platform) were the technologies used to bear traffic between 2G/3G base stations to the core network.
  • Synchonization is required in time division multiplexing and it was complex to implement.
  • TDM was also bandwidth inefficient, remember ISDN/PSTN telephone lines, where wired lines were dedicated for each user ? Each E1/T1 interfaces BW rate is relatively low and cost is high.
  • TDM technologies as huge Operation Expenditure (OPEX) cost. Field engineers needed to be on site to troubleshoot the circuit test (which is loop test) to validate the connectivity.

High Level Designing

Underlay - IGP

IS-IS protocol is the winner for the IGP because of its fast convergence and link metric planning. Aggregation ring is configured as a base area as access ring will converge to aggregation.

ISIS 1
router-id <X.X.X.X>
ISIS 100
 router-id <X.X.X.X>
#
interface GigabitEthernetX/X/X
ISIS enable <XX>
 ISIS cost <cost value>
ISIS authentication-mode md5 1 cipher <XX>
 ISIS circuit-type p2p
 ISIS ldp-sync
 ISIS timer ldp-sync hold-max-cost XX 
#
isis 1
is-level level-2
cost-style wide
timer lsp-generation 1 50 50 level-2
flash-flood 15 level-2
 bfd all-interfaces enable
 bfd all-interfaces min-tx-interval 100 min-rx-interval 100
 is-name <description>
 network-entity <49.XXXX.XXXX.XXXX.XXXX.XX>
 timer spf 1 50 50
 traffic-eng level-2
 log-peer-Change
 set-overload on-startup
"""

MPLS

It involves various design on MPLS-LDP and MPLS RSVP-TE.

MPLS LSR-ID

#
interface Loopback0
  description LSR-ID
  ip address 10.X.X.X
#
mpls lsr-id 10.X.X.X
#

LDP Design

The LDP works on hop-by-hop basis to form LSPs, which follow the shortest path from ingress to egress as selected by the IGP. An LSR running LDP discovers its neighbors and automatically distributes a label for each address for which it has received a label.

LDP LSPs’ quantity relevant to IGP’s /32 mask prefixes amount by default, while LDP label withdraw and session re-built also determined by IGP fast convergence. By using ISIS stub area per access ring scheme to maintain ISIS routing table at light-level is one sensational approach. In this way, the LSR only advertise labels to the necessary destinations. Small IGP routing/LDP label table is agile when experiencing topology changes.

Configuration example:

#
mpls ldp
#
interface GigabitEthernet1/1/0
mpls     
mpls ldp 
#

Traffic Engineering

RSVP Tunnel

Traffic engineering (TE) allows the network operator to have better control directing traffic along a specified path rather than the best IGP path. When deciding the traffic’s route, the system can consider aspects (etc. explicit path, administrative group.) other than the IGP cost. Constrained Shortest Path First (CSPF) is the algorithm enhanced from the SPF to take TE information into account when calculating a path. The TE extension needs to be enabled in the IGP view before the deployment of TE tunnels. TE and CSPF will use together to calculate the path for the MPLS RSVP-TE LSP.

Configuration Example:

mpls
mpls te                 
mpls rsvp-te 
mpls te cspf            
#
interface GigabitEthernet2/1/0
mpls te 
mpls rsvp-te                
#

MP-BGP

IPRAN deploys MP-BGP to provide reachability for VPNv4 unicast prefixes for the IP RAN solution. The MP-BGP deployment scope covers all the IP RAN devices from PE,CE,ASG and CSG nodes. BGP is deployed on all routers to support L3VPN services. The master and slave ASGs establish BGP VPNv4 peer relationships with the master and slave RSGs and CSGs respectively. To support transmission over the shortes path in the LDP-LSP scenario, ASGs are configured as route-reflectors.

#
Bgp XXXX
ipv4-family vpnv4 
 peer X.X.X.1 route-policy rp_from-rsg import
  peer X.X.X.1 route-policy to-rsg export
  peer X.X.X.1 advertise-community
  peer X.X.X.2 route-policy rp_from-rsg import
#

Route-Policy: Import

#
route-policy rp_from-rsg permit node 10
 if-match community-filter 10
 apply preferred-value 5
route-policy rp_from-rsg permit node 20
 if-match community-filter 20
 apply preferred-value 0
route-policy rp_from-rsg permit node 100
ip community-filter 10 permit 65004:1111 
ip community-filter 20 permit 65004:2222
#

Route-Policy: Export

#
route-policy to-rsg permit node 10
if-match ip-prefix nodefault
ip ip-prefix nodefault index 10 deny 0.0.0.0 0
ip ip-prefix nodefault index 20 permit 0.0.0.0 0 less-equal 32
#

VPN Route Policy (eg: 2G service)

ipv4-family vpn-instance Service_2G
  network 0.0.0.0 route-policy preferred-value  
  import-route direct route-policy preferred-value 
route-policy preferred-value permit node 10
 apply preferred-value 32768

Route Reflectors

To avoid full-mesh BGP peering, Route Reflectors (RR) will be used. Those paired RSG will serve as sub-region RR reflecting both VPNv4 and Unicast routes to all the CTs and ASGs. RR RSGs and their client CTs, ASGs are divided into clusters according to sub-region info by geographical location for network scalability. The following principles will be applied to the RR’s design and homing relationship:

  • RRs will use the router id as cluster id, instead manual set with identical cluster id.
  • All ASG will be reflected as a Route Reflector for their respective CSG ring nodes and RSGs and both RSG will act as a route reflector for their respective ASGs.
  • All the routers each Access Ring should establish IBGP peers with two RRs separately.
  • Build full meshed IBGP session among ASGs.
  • Use peer groups with RRs to reduce configuration complexity and improve convergence time.

For RSG/RR sites:

bgp 65000
#
ipv4-family vpnv4
  reflector cluster-id 100
  reflect change-path-attribute
peer ASG reflect-client
#

The Big Picture

Multiple IP-RAN networks converges to IP-CORE and Data-Center.

/images/blogs_images/ipran/ipran-2.svg