At a glance, veterans will probably know that it must be about the introduction of common solutions such as "link aggregation", "spanning tree", "dynamic routing", and "loop protection". Haha, it's not that simple, you're half right, but what I'm going to share today is a set of plans! Those who follow me should know that I am a "dry goods gentleman", and I believe that you will be very interested in this plan~
Let's first introduce the most common redundant backup technologies in several networks:
1. Common redundancy techniques
【Port Aggregation】
Port aggregation is to aggregate multiple physical ports of a switch together to form a logical physical port, and multiple links in the same aggregation group can be regarded as one logical link. Port aggregation has two main functions: bandwidth stacking and link backup.
【Spanning Tree】
Spanning Tree Preventive (STP) is used to eliminate link-layer physical loops in a LAN and automatically activate the backup link recovery network when a link fails. Devices running this protocol discover loops in the network by exchanging information with each other, and selectively block certain ports, and finally prune the loop network structure into a loop-free tree network structure. The protocol types are STP, RSTP, AND MSTP.
【ERPS】
I call it the best loop-breaking protocol, with a convergence time of less than 50ms! Switches in industrial environments need to operate reliably, even if one device fails or is disconnected, it can be replaced by other devices, and the switching is fast and without delay. The industrial-grade ERP ring network can meet this use scenario, which is faster than the general spanning tree protocol switching, and perfectly achieves non-interrupted business switching.
When the ERPS ring network is working normally, the RPL port blocks the link where it is located to prevent network loops. When a link in the ring network is abnormally disconnected, the abnormal node will send a packet to notify the RPL owner node, so that the RPL port is released, so that the link where it is located can resume normal communication, and ensure that the communication of the entire network is not interrupted. There is only one RPL Owner node in a ring network.
2. Introduction to wireless redundancy scheme
The following is the main content of this issue, combined with the above redundancy technology, I recommend a set of wireless redundancy solutions to you:
This is a set of wireless link redundancy solutions that are simplified from the industrial network environment!
【Solution Requirements】
- The network is designed according to loop redundancy, and the standby link for primary link failure is enabled to ensure rapid communication recovery.
- When the active link recovers from the fault, the standby link switches to a blocked state, and the primary link is responsible for communication.
- Save cabling costs as much as possible to achieve link backup, and ensure high-throughput forwarding for wireless;
- The data forwarding of the wireless link must be consistent with the active link, including the processing of VLAN tag data stream and MAC address table.
【Solution Technology】
Choose one of the following to achieve link redundancy, do not use both!
Port aggregation: The access switch and the redundant switch converge the active and standby link ports to implement redundant loads.
Spanning tree: RSTP is enabled on the primary/standby link ports of access switches and redundant switches to implement loop redundancy.
ERPS: The ERP function is enabled on both the access switch and the redundant switch, and the standby link port of the redundant switch is configured as an RPL port.
【Solution Implementation】
- Network cables or optical fibers are deployed on the main link between the access switch and the redundant switch, and the backup link uses wireless data transmission to achieve data transparent.
- Redundant switches and the core layer can be redundant through aggregation and RSTP.
- The standby wireless link is implemented by using a wireless bridge (or industrial-grade wireless AP + wireless client, similar to the bridge function).
- Considering whether the standard is for transparent transmission of data flows with VLAN tags, the wireless bridge generally needs to be configured to fully transparently transmit all VLAN data, or configure the specified VLAN tag data for transparent transmission.
- The bridge adopts 5GHz frequency band products to ensure high-speed and high-stability wireless forwarding;
- The bridge wireless address format is required to be "four addresses" to prevent errors in the MAC address table of the learning subnet on the main bridge side.
- WDS wireless bridging has two formats: three-address and four-address, and the main bridge (AP) cannot learn the real mac of the wired device under the sub-bridge (client) when communicating with three addresses:
- With four-address format communication, the AP side of the main bridge can learn the real mac of the wired device under the subnet bridge (client):
3. Implement wireless link redundancy in different ways
Solution 1: Port aggregation implements link load
- The port aggregation traffic will go from both the primary link (wired) and the standby link (wireless), which has the advantage of achieving bandwidth superposition to a certain extent, but the disadvantage is that when the wireless link quality is poor, some data communication will produce delay and packet loss.
- The port aggregation function must be enabled on the two ports interconnected between the redundant switch and the access switch, and the LACP mode must be configured instead of the static aggregation mode.
- Note: Static aggregation relies on link state awareness, and if port 1 of the access switch fails, all traffic will go through port 2 for the access switch. However, if ports 1 and 2 of the redundant switch are connected normally, traffic may still go out through the two ports of the redundant switch, which will cause packet loss.
Solution 2: Spanning trees implement link redundancy
- The redundant switch is configured as a root bridge, and the access switch is a non-root bridge, and RSTP or MSTP is enabled.
- The cost value (path cost) of the 2 ports of the non-root bridge access switch must be larger than that of the 1 port, so that the 2 ports will be selected as the AP port blocking as the backup link.
- After the fault, the AP port becomes the RP role, the status is forwarding, and data forwarding is resumed on top of the standby link. Once the active link is restored, port 2 of the access switch becomes blocked by AP port.
Solution 3: ERPS implements link redundancy
- Port 2 of the redundant switch is configured as an RPL port, and the other ports and the port of the access switch are normal interfaces.
- If the link is normal, the RPL state is blocked, and the primary link is responsible for data forwarding. After the primary link is abnormal, the RPL is set to the forwarding state, and the standby link is on top. After the primary link is recovered, after a period of time, the RPL switches back to blocking and resumes the forwarding of the primary link. The convergence time of active/standby switchover and fault recovery is less than 50 ms, achieving "seamless switchover" in the true sense.
- The wireless bridge must support transparent transmission of data from ERP control VLANs and service VLANs.
Fourth, product options reference
Combined with the above, to realize this solution, the focus is on the selection of wireless bridges and switches, as follows:
Switchboard:
- 支持STP/RSTP/MSTP；
- It supports port aggregation and LACP mode.
- Support ERP function (basic function of industrial-grade switches)
Wireless Bridge:
- Support four-address wireless format bridging (basically on the market)
- Transparent transmission of VLAN data is supported.
- Supports transparent transmission of Layer 2 multicast and broadcast packets of protocols such as BPDU, lacp, and CCM (i.e., to ensure the normal use of aggregation, spanning tree, and ERPS functions)
- Choose 5GHz bridge products whenever possible.
Regarding the bridge, I have personally tested Ruijie's RG-EST310, TP-LINK's TL-CPE520G and Mercury's B5 suit, which can basically meet the above basic requirements.
This solution comes from the industrial wireless I have been exposed to, and using industrial AP+industrial client is also a feasible solution, but the cost will be higher. In practice, a normal wireless bridge will do.
Comrades who are interested in this plan can discuss with me, leave a message in the background or add me to serve WeChat, and I can receive the message