Table of Contents |
---|
Hide_comments |
---|
Abbreviations
The following abbreviations are used in this document:
- IP – Internet Protocol
- MAC – Media Access Control
- MINT - Mesh Interconnection Networking Technology
- PRF – Pseudo Radio interface
- RCMD – Remote Command
Purpose of this document
InfiNet Wireless is one of the leading manufacturers of Broadband Wireless Access equipment for carrier grade fixed installations. InfiNet Wireless uses its own proprietary transport protocol – MINT which interconnects units by wireless and wired (MINT-over-Ethernet technology provides MINT connectivity over wired Ethernet) links. Within MINT areas, it is possible to send any command via the MINT protocol to a specified unit for execution (MAC address can be used for specific unit selection or broadcast MAC addresses can also be used).
This document shows how units management can be organized using MINT protocol for InfiLINK 2x2 and InfiMAN 2x2 families devices. MINT protocol operates on both Layer 2 and Layer 3. However, in case incorrect or missing IP settings on the units, the management configuration on InfiNet Wireless unit can be restored by issue of MINT RCMD (remote commands) from another InfiNet Wireless unit. MINT management only required L2 and MINT connectivity between units.
Introduction to MINT protocol
The MINT main purpose is to provide path selection with best quality for wireless (and wired) traffic on Layer 2 (switched traffic).
MINT highlights
Path quality check | MINT constantly checks transmission quality for each link. In case of link degradation MINT quickly changes some parameters in order to keep packet loss value as low as possible automatically. In case of redundant links available, MINT will switch main traffic flow path to link with better transmission quality. |
Predictive model | MINT supports predictive model to select the best path in advance, in order to deliver data as quickly as possible. |
Redundancy and load balancing | MINT protocol was designed to work with multiple redundant paths. Moreover, such redundancy can be used for load-balancing to utilize all available connections from one customer point to another. |
Minimum time for data delivery | MINT main criteria for optimal path selection is time (minimal packet delivery time). |
MINT connections via wired Ethernet (MINT-over-Ethernet) | MINT-over-Ethernet is unique feature of IW, allows to select best path through network including wired interfaces. Thus, both wireless and wired interfaces would be utilized. MINT-over-Ethernet can be enabled to provide backup and redundant paths, especially for mobile projects. Pseudo Radio Interface (PRF) are virtual interfaces created on base of Ethernet interfaces provides MINT-over-Ethernet. |
Switching loop prevention | MINT has built-in mechanism to prevent data from looping within MINT network (when redundant paths are present). |
MINT position
MINT link is the link between two units, which use MINT as transport protocol.
MINT encapsulates and transports all traffic and all protocols. Thus, MINT is the only one transport protocol for InfiNet Wireless InfiLINK 2x2 and InfiMAN 2x2 families.
MINT operates between DataLink and Network Layers of the OSI Model. Therefore MINT is capable to encapsulate and carry through link Layer 2 traffic (Ethernet switched data) and Layer 3 traffic (IP routed data).
Center | ||||||
---|---|---|---|---|---|---|
|
MINT technology overview
Where MINT is available
Only adjacent neighbor connectivity is needed to create MINT powered link:
- For radio interfaces MINT is enabled by default, only radio parameters should be configured;
- For Ethernet interfaces MINT-over-Ethernet should be enabled manually (disabled by default) for Ethernet interfaces. In case to perform it, Pseudo-radio interface (PRF) should be created. Each BS secor and CPE supports such interfaces.
MINT path selection
In case of multiple MINT routes the protocol would always choose one route for single data frame and would have possibility to re-select new route for another data frame in case of any path characteristics change. Path characteristics are described by aggregated parameter MINT cost.
MINT cost is calculated from the following parameters:
- Signal-to-Noise Ratio (for connectivity over radio interfaces);
- Throughput or Bitrate (for connectivity over radio interfaces);
- Percentage of retries ;
- Link load and throughput;
- Some other parameters.
Link quality assessment:
- Each MINT unit has full MINT map with all MINT neighbors;
- Each MINT neighbor constantly checks MINTcost between each other;
- Checking period depends on mode selected: for Fixed mode - 3s, Nomadic - 1,5s, Mobile - 1s;
- MINT path can be predictably changed due to change in link quality (cost drop).
Loop free capability:
- The path for data frame is selected by the lowest overall cost;
- The predefined path is set for every packet, unless any changes in network;
- The path for frame or packet can be different;
STP BPDU transmission can be blocked in configuration through any logical interfaces.
MINT path selection capabilities
MINT protocol will quickly adjust to possible changes in critical parameters – re-calculate MINT cost for each path and rebuild the path accordingly. Moreover, due to built-in capability to quickly adopt changes, even data flow path within MINT network can change rapidly too.
On the picture below shown mesh topology of units. All units run MINT protocol for each interface, hence each connection is handled by MINT protocol. In this case, node J has to send data to node F.
Center |
---|
The path selection process are detaily shown in the video below:
Center | ||||
---|---|---|---|---|
|
MINT area
All benefits of MINT protocol mentioned above, would be active only in network there all units supports MINT protocol as the only one transport protocol. Such MINT network is called MINT area. Thus it is required to create unified MINT area consisting of InfiNet Wireless units interconnected by wireless radio interfaces (RF) or by wired MINT-over-Ethernet (Pseudo Radio) interfaces.
Center |
---|
So, here we have network where almost each unit has at least two connections via MINT protocol. Therefore, it is possible to balance traffic via one or second path (or even load balance using both). In case of one link failure, traffic will flow through the another one. Eventually, under certain circumstances the units can always stay connected because both links down situation is very unlikely to happen. It doesn’t really matter what type of physical connection is used (wired or wireless), MINT use any connection. The only difference is MINT cost value for each link.
Note | ||
---|---|---|
| ||
For more information about traffic balancing proceed to article Link aggregation, balancing and redundancy. |
MINT RCMD
Within MINT area every MINT node can receive information about another MINT node through its MINT neighbors. Every MINT neighbor exchanges information about its adjacent links, their quality, load, issue and MAC address. MAC address is used as identification label for MINT node.
List of MINT neighbors:
Code Block | ||||
---|---|---|---|---|
| ||||
Master#>console>mint map detail ============================================================================== Interface rf5.0 TDM (5 ms DL/UL:Auto) (RSSI=-43 Dist=1) (Sync Off) Node 00043513724F "Master", Id 25871, Nid 0, (Master) Freq 5470, Band 20, Sid 10101010, autoBitrate 130000 (min 13000), Noise -94 ------- ---------------------------- ------------ ----- ------- ----- ------- Id Name Node Level Bitrate Retry Options ------- ---------------------------- ------------ rx/tx rx/tx rx/tx ------- 60755 Slave 00043523FA93 14/14 104/104 0/0 /S/ load 0/4, pps 0/1, cost 51 pwr 25/27, rssi -57/-53, snr 33/36 dist 0.14 H11v2.1.11, IP=10.10.10.2, up 18 days 60756 Slave 2 00043523FA94 18/30 117/130 0/0 /S/ load 5/2, pps 2/0, cost 51 pwr 12.5/27, rssi -50/-21, snr 41/63 dist 0.18 H11v2.1.11, IP=192.168.103.37, up 14 days ------- ---------------------------- ------------ ----- ------- ----- ------- 2 active neighbors Total load: 5/6 (rx/tx), 11 (sum) Kbps Total nodes in area: 3 |
So, MINT neighbor is designated by MAC address.
Unique feature of MINT protocol is possibility to send any command to any MINT neighbor for execution at MINT neighbor unit. It is called MINT Remote Command execution (RCMD). MINT RCMD could be helpful in lots of cases, such as: lost password, ip address settings cleared, no possibility to login to remote MINT neighbor directly, it is required to execute some commands on all MINT neighbors.
Examples
Force reboot on remote MINT neighbor
| |||||||
Set new IP address and Default Gateway on remote MINT neighbor
|
Note | ||
---|---|---|
| ||
Full syntax of MINT remote commands with different options is described in WANFleX command reference guide Layer 2 commands set -PHY and MAC. |
MINT area prerequisites
In default configuration MINT protocol is enabled and used only between wireless radio interfaces. However, in order to create interconnected MINT area MINT-over-Ethernet interfaces (PRF) are required.
In order to enable MINT-over-Ethernet proceed to the following:
- Create virtual Pseudo Radio Interface (PRF). PRF can be created as logical sub-interface for physical Ethernet interface (parent interface), or as logical sub-interface for another logical interface;
- Start MINT protocol for PRF interface;
- JOIN command creates internal connection between Radio interface and PRF interfaces.
Within the same Ethernet broadcast domain (LAN) two (or more) InfiNet Wireless units with PRF interfaces created, can find each other and establish connection via MINT protocol by sending and receiving Ethernet broadcasts frames. Thereafter, InfiNet Wireless units will use Ethernet unicast data transfer.
VLAN considerations
In case two InfiNet Wireless devices are placed within certain VLAN, then configuration of PRF should be performed with certain VLAN tag specified.
MINT-over-Ethernet & VLANs
MINT-over-Ethernet generates broadcast traffic to find other MINT neighbors. Sometimes, such broadcast traffic could be treated as abnormal for network, especially for enterprise networking with comprehensive network security policy.
Note | ||
---|---|---|
| ||
It is recommended to put MINT-over-Ethernet connections into dedicated VLAN zones, thus keeping all broadcast traffic within unique VLAN |
Moreover, in whole MINT-over-Ethernet network different MINT-over-Ethernet areas should be isolated from each other in order to provide complex traffic engineering or prevent undesired traffic path selection. Hence, VLAN separation should be used in such cases.
Configuration part
Steps to configure MINT-over-Ethernet are shown below. Only step 4 is different for untagged approach (4b) and for vlan based approach (4a).
CLI based configuration
1. Create Pseudo Radio (PRF) interface on all devices and set eth0 as parent:
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
ifconfig prf0 up prf0 link administratively up Master#console>prf 0 parent eth0 OK. |
2. Start MINT protocol for PRF interface:
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
mint prf0 start OK. |
Check the connection between devices:
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
mint prf0 map detail ============================================================================== Interface prf0 (parent eth0) Node 00043503724F "Master", Id 25871, Nid 0, (Master) ------- ---------------------------- ------------ ------- Id Name Node Options ------- ---------------------------- ------------ ------- 60756 Slave 2 00043503FA94 prf load 64/0, pps 6/0, cost 51 H11v2.1.11, up 00:00:24 ------- ---------------------------- ------------ ------- 1 active neighbors Total load: 64/0 (rx/tx), 64 (sum) Kbps Total nodes in area: 2 |
Right now we have created and enabled radio link and backup MINT-over-Ethernet paths. We have just created and enabled TWO independent instances of MINT protocol. Both instances are completely independent and have no information of each other, no matter they are started inside the same unit.
Therefore, the next step is to join these instances together.
3. Make unified MINT interface by join command:
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
mint join prf0 rf5.0 Complete list of joined interfaces: mint join rf5.0 prf0 |
Now, both MINT interfaces are treated by the unit as single entity, hence MINT exchange information using both legs (wireless and wired).
4. In order to allow IP management traffic reach the unit, switch group need to be configured.
4a. In case of a VLAN based management, a VLAN interface is requiered with parent eth0 interface. Created VLAN interface and one of joined interfaces (rf or prf) need to be added to the switch group. We strongly recommend you to set a switch group number same with the VLAN used to avoid confusion.
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
ifconfig vlan100 vlan 100 vlandev eth0 up sw group 100 add vlan100 rf5.0 OK. sw group 100 start OK. |
4b. If no VLAN is used for management, it is enough to add eth0 interface and one of joined interfaces (rf or prf) to the switch group.
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
sw group 100 add eth0 rf5.0 OK. sw group 100 start OK. |
5. Create management svi interface, add it to the the group and set the IP address to svi interface.
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
ifc svi100 up svi100 link administratively up svi 100 group 100 OK. ifc svi100 192.168.1.3/24 OK. |
6. Don't forget to save the configuration.
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
config save |