Question Can unmanaged switches do "passive" LACP? Is that possible?

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,202
126
Just ordered one of these bad boys, for $160 + ship + tax.

Netgear, 8x 1Gbe, 2x 10GbE, unmanaged

https://www.amazon.com/NETGEAR-Ethernet-Unmanaged-Multi-Gig-Protection/dp/B076642YPN?th=1

They offer a "managed" version for $250.

Is it possible, if I configure my QNAP NAS units for Teaming / LACP, that the unmanaged switch will automagically configure the two ports that they are plugged into for LACP as well? Is "Passive LACP" a thing? Or do switches NEED to be "managed" to be able to create LACP groups?

I know that with routers, there was "WDS" and "lazy WDS". Lazy WDS was passive, and didn't require configuration on the destination end, to initiate a WDS link. My hope is that LACP, however it actually works, can do the same thing. Or am I just ignorant to the specifics? This is the first time that I've messed with this feature.

PS. THANK YOU, mystery admirer, for the Amazon GC. You know who you are. (But I don't.)

Practical notes
LACP works by sending frames (LACPDUs) down all links that have the protocol enabled. If it finds a device on the other end of the link that also has LACP enabled, it will also independently send frames along the same links enabling the two units to detect multiple links between themselves and then combine them into a single logical link. LACP can be configured in one of two modes: active or passive. In active mode it will always send frames along the configured links. In passive mode, however, it acts as "speak when spoken to", and therefore can be used as a way of controlling accidental loops (as long as the other device is in active mode).[4]
https://en.wikipedia.org/wiki/Link_aggregation#cite_note-ieee802.org-4

Thank you Wikipedia!

Edit: According to the datasheets, the unmanaged one can't even do passive LACP. Seems unfair to me. Sounds like product segmentation if anything.

https://www.netgear.com/images/datasheet/switches/GS110EMX_GS110MX_DS.pdf


Edit: Now I'm wondering, if I should have bought the TrendNet 24x 1GbE + 4x 10GbE SFP+ web managed switch for $279.

https://www.newegg.com/Product/Product.aspx?Item=9SIA0ZX4FE5652

https://www.cnet.com/products/netgear-gs110mx-switch-10-ports-unmanaged-rack-mountable/

CNet specs state:
Features
128KB packet buffer,
DDos attack prevention,
DiffServ Code Point (DSCP) support,
Energy-Efficient Ethernet (EEE),
IGMP snooping,
Link Aggregation,
Weighted Round Robin (WRR),
cable test,
fanless,
flow control,
full duplex capability,
port mirroring,
security lock slot (cable lock sold separately)

(Bolded mine)

According to CNet, it CAN do LAG, but they also mention Port Mirroring, which I didn't think was possible to do passively, I thought that required some sort of mgmt layer to configure that.

So maybe they're mixing the specs up between the two switches, the MX and the EMX (managed).

I guess I'll find out later today.

https://www.smallnetbuilder.com/lanwan/lanwan-howto/30556-how-to-set-up-switch-link-aggregation

A final interesting aspect about a LAG is it doesn't increase throughput for individual data flows. Each data flow is limited to the bandwidth of a single link in the LAG. In a LAG with two or more 1 Gbps links, the best throughput an individual data flow will see is 1 Gbps. The real value of LAG is in increasing total (or aggregate) throughput between devices. Read this brief presentation for a nice and clear explanation.

I thought LAG/LACP effectively bonded two ethernet ports into one bigger aggregate channel. No?

This link shows an example of link aggregation, using a Netgear switch, that is "unmanaged", but using a separate app on a PC to configure it?

https://www.pcgamer.com/how-to-set-up-nic-teaming-link-aggregation/
 
Last edited:

thecoolnessrune

Diamond Member
Jun 8, 2005
9,673
583
126
CNET is a poor place to get specs, and they're very often wrong. In this case, Netgear's Datasheet gives you the information you need: https://www.netgear.com/images/datasheet/switches/GS110EMX_GS110MX_DS.pdf

The Unmanaged switch does not do any trunking. The GS110EMX does is the Managed version that does do trunking.

LAGs also do not multiply bandwidth for a single stream. LAGs add lanes to a highway, it doesn't make the lane wider. You have to move up to the Protocol level to get traffic broken up in a way that a single application flow can utilize all the lanes on the highway. SMB3's multichannel for instance can do this. So can NFS4.1's pNFS specification.
 

Genx87

Lifer
Apr 8, 2002
41,091
513
126
I'm sure the QNAP is like Synology in they have a dumb teaming mode that will work with unmanaged switches. Look on ebay, you will find decent managed switches that support LACP for $80-$150 bucks.
 

mxnerd

Diamond Member
Jul 6, 2007
6,799
1,103
126
If your QNAP NAS supports SMB 3.0 or you have a Windows Server 2012 and up and your client machines are Windows 8.x/10 PCs, then you don't need a LACP capable managed switch.

SMB 3.0 supports multichannel, which can combines multiple gigabit ports.

==

Quadruple Your Network Speed for $100 with SMB 3.0 Multichannel! (** Machine's CPU matters **)


==

Btw, configure your LACP capable QNAP & switch

 
Last edited:

JackMDS

Elite Member
Super Moderator
Oct 25, 1999
29,528
415
126
Everything is "good and dandy".

The only thing that is lacking is an explanation to why you need this switch?

This switch is a component for Server Farm and Not Home network.


:cool:
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,202
126
This switch is a component for Server Farm and Not Home network.

Well, I do have kind of a mini server farm.
Have several (many?) NAS units, some of which have multiple 1GbE-T ports. Hope to get a NAS with a 10GbE-T port, and a 10GbE-T NIC for the "main" PC. (Or possibly just a new AM4 mobo with 10GbE.)
 

thecoolnessrune

Diamond Member
Jun 8, 2005
9,673
583
126
I assume the Synology "Adaptive Load Balancing" is a variant of "FLAG" (Fake LAG). It will use a technique similar to ARP poisoning where replies from the Synology are over-written by its bonding driver with the MAC Address of various interfaces in the bond, chosen by current load. This will have responses back to the Synology be destined for that forged MAC Address. This creates a wave effect on traffic received by the Synology. Some data will come on the non-primary interface, until another ARP request is broadcast. At that point, each client receives the real MAC Address of the bond and all traffic collapses back on one link. That is until the unit starts receiving the packets and re-writing the ARP responses again. This process works pretty well in static home environments, but noisier environments with more ARP broadcasts can minimize the benefits. The premise behind this of course is the belief that environments small enough to be unable to use LACP would also be small enough to minimize the need for constant ARP updates.