3

Once here I ask about LACP between rb750 and huawei ne20, as @Zac67 said

"throughput for multiple flows, not for a single flow"

. I was testing from a single flow, now is working, but I came across with something after this.

My new test is: establish 2GB between a Huawei Switch s6720 and Mikrotik Rb1100. As you can note, before I was using Huawei Ne20.

The biggest different that I noted was a single configuration about LACP MODE

 ## NE20 ##
 interface Eth-Trunk1
 portswitch
 port link-type access
 port default vlan 200
 mode lacp-static

 ## Switch S6720 ##
 interface Eth-Trunk1
 port link-type access
 port default vlan 200
 mode lacp

Display Information Huawei S6720

Eth-Trunk1 current state : UP
Line protocol current state : UP
Description:
Switch Port, Link-type : access(configured),
PVID :  200, Hash arithmetic : According to SIP-XOR-DIP,Maximal BW: 2G, Current BW: 2G, The Maximum Frame Length is 9216
IP Sending Frames' Format is PKTFMT_ETHNT_2, Hardware address is f098-3827-8570


Local:
LAG ID: 1                   WorkingMode: LACP
Preempt Delay: Disabled     Hash arithmetic: According to SIP-XOR-DIP
System Priority: 32768      System ID: f098-3827-8570
Least Active-linknumber: 1  Max Active-linknumber: 8
Operate status: up          Number Of Up Port In Trunk: 2


Now, I'd like to compare their display Eth-Trunk 1 (S6720 and Ne20). Above is s6720, bellow is Ne20.

Local:
LAG ID: 1                       WorkingMode: STATIC
Preempt Delay: Disabled         Hash arithmetic: According to flow
System Priority: 32768          System ID: 487b-6b4a-cc5c
Least Active-linknumber: 1      Max Active-linknumber: 16
Operate status: down            Number Of Up Ports In Trunk: 0
Timeout Period: Slow


Pay attention to WorkingMode: one is Static and the other just LACP

Testing LACP (Using BTest)

Ne20 Aggregation:

enter image description here

Both Mikrotik (A,B,C) are sending 1GB to (Mikrotik Port-Aggregation). That scenario is working.

enter image description here

Same test using btest, but in this scenario, just 1GB is coming.

enter image description here

I'm not sure but I think the problem is the mode which should be "static" but S6720 does not allow this mode.

  • Anyone has experience with RouterOS ? – TMoraes May 21 '18 at 19:18
  • Did any answer help you? If so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you can provide and accept your own answer. – Ron Maupin Dec 25 '18 at 8:27
5

Static LAG trunks are just that: static. When multiple interfaces are connected they are regarded as a single logical port.

LACP trunks are negotiated by protocol. That way you can ensure that they are terminated correctly (each LACP group can only be connected between two switches). Incorrectly terminated interfaces are not added to the LAG trunk. You should prefer LACP over static in any possible case.

An LACP trunk on one side will NOT work with a static trunk on the other side. Likely, the static side will aggregate the link while the LACP side won't, resulting in rather unpredictable behavior, MAC flapping and such.

LACP comes in two modes, active and passive. A passive interface will accept an LACP connection but will not initiate it. An active interface does both. I use active LACP at all times.

I'm not sure what problem you're trying to solve. Once established, static and LACP trunks behave the same way. Usually, traffic is split across the physical links using SA/DA hashes: depending on the device and its configuration, flows between the same MAC addresses or IP addresses or IP&TCP/UDP port combination will always use the same physical link.

If the switches use MAC SA/DA all traffic between them will always use the same port combination. Very much the same goes for IP addresses (but you might work around it using multiple IP addresses). With TCP/UDP SA/DA each transport layer flow will stick to the same ports, but multiple flows may use different ports, leveraging the LAG group.

However, each SA/DA combination results in a somewhat random port index in the trunk. With few flows, you will not get an even distribution unless you're very lucky or fine-tune the addresses.

For testing you need to send many different flows through the trunk and check whether the traffic is somewhat distributed.

1

Without having any experience with either system families, I'd recommend that you seek out what the manufacturer means with just "LCAP", the terminology for LCAP settings vary wildly between manufacturers. LCAP-static is a bit weird in my mind, either you have a static LAG or a dynamic LAG (LCAP for example). Set both sides to either a static setting, or the right LCAP setting.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.