Tag Info

Hot answers tagged

8

What you're looking for is commonly called a "transmit hash policy" or "transmit hash algorithm". It controls the selection of a port from a group of aggregate ports with which to transmit a frame. Getting my hands on the 802.3ad standard has proven difficult because I'm not willing to spend money on it. Having said that, I've been able to glean some ...


8

The document on Linux bonding is worth the read, it goes into what sort of support you need from switches for various bonding methods. In this case: The balance-rr, balance-xor and broadcast modes generally require that the switch have the appropriate ports grouped together. The nomenclature for such a group differs between switches, it may be ...


7

LACP itself doesn't provide the ability to bond across multiple switches; it bonds across multiple ports on a single ethernet switch, and depending on the vendor there might even be restrictions on which ports on a switch can be bonded together. Some vendors have proprietary protocols (typically called MLAG) that allow for bonded ethernet channels across ...


7

That's normal. With a single stream of data over an LACP bond, only 1 connection will be used. LACP is useful when you have multiple data streams traversing the link and for active/active connection redundancy. Try copying multiple large files to/from different destinations and the utilization should reach close to 100% if it is configured correctly.


6

The two main benefits to distributed switches are More features. LACP as you mentioned Visibility into the network activity on each virtual port (so you can see the unicast/multicast/broadcast counters for a specific VM in the vCenter interface) CDP advertisements from the vDS to the physical network devices Mirroring/SPAN for monitoring or ...


5

Both ends need it setup right.


4

Failover: Unplug each cable and ensure traffic continues to flow between the switches. Bandwidth: iperf seems to be the standard took for testing these kinds of things. If you can get a setup like this: A ----- |--------|===L1===|---------| ------ C | Switch | | Switch2 | B ----- |________|===L2===|_________| Then setup Iperf on ...


4

More likely than not the teams are using LACP for increased throughput and failover. If this is the case manually creating an EtherChannel is redundant and will lock the port group into specific manually configured ports (rather than LACP's dynamic assignment of ports). Basically it sounds like the configuration is 100% correct and you shouldn't mess with ...


4

To be honest, my view of this is that purposely designing a loop in your network design is not a good design. Spanning tree can be a major pain point to manage, design, implement, troubleshoot, etc. LACP and STP are two completely different things. At a very high level, LACP is what lets you create your LAG - it will take multiple interfaces and treat them ...


4

Distributed Trunking was a means on the ProVision family of HP Switches (35xx/54xx/62xx/66xx/82xx.. not 38xx) to be seen as one switch. It was still two configurations, and there are some features that get disabled on the ProVision switches if enabled. When Distributed Trunking first came out, it was Server to Switch. Its since been improved from switch to ...


4

A quick and dirty explanation is that a single line of communication using LACP will not split packets over multiple interfaces. For example, if you have a single TCP connection streaming packets from HostA to HostB it will not span interfaces to send those packets. I've been looking at LACP a lot here lately for a solution we are working on and this is a ...


4

LACP will not split packets across multiple interfaces for a single stream/thread. For example a single TCP stream will always send/receive packets on the same NIC. See the following post for reference: Link aggregation (LACP/802.3ad) max throughput Hope this helps.


3

LACP aggregates multiple physical links between the same two devices into one logical link that has higher throughput. If you are looking for redundancy in case of a switch failure, then you need to set up links to two switches, and configure the server to bridge the two links with the spanning tree protocol enabled. STP will automatically choose one link ...


3

LACP is a protocol for the host to communicate to the switch(es) that it wants to aggregate multiple physical connections. This aggregation might not always be possible however. If the connection is made to multiple switches those switches must support some method of cooperating to make the bonded link or they will be unable to. Most switches do not ...


3

LACP can usually only be accomplished to a single device or group of device that act as one. So... you could do it between a switch and a switch stack, but not spread out across multiple unique devices.


3

You can split a bond across the switches, but not to increase throughput. So if a server has two NICs and you plug one into each Switch, then you'll have to configure the server for some sort of bonding with failover only. This Q/A has a lot of good details: Server-to-Switch Trunking in Procurve switch, what does this mean? The switches have to be plugged ...


3

In balance-alb, both sending and receiving frames are load balanced using the change MAC address trick. This might cause issues at application levels. Not all applications are matured for this mode. To Handle your original issue. Here is what I used to do. Leave the switch ports to default. Perform pxe-Kickstart installation. Either at KS post ...


3

I'm not terribly familiar with Juniper switches, but you shouldn't have to configure LACP on them; that is the point of LACP. If this isn't the case, something is wrong with your switch configuration. LACP only specifies a protocol for dynamically aggregating ports. It does not specify a port scheduling policy (where traffic is sent and received). This ...


3

Yes, and no. In the "usual" modes for LACP aggregation, which link a particular packet will flow down is determined entirely by either the source or dest MAC (or both, if hashed). So you are correct in saying that the most traffic that a single MAC address (or src+dest pair) can do is the speed of a single one of your links. However, if a link is ...


3

From my experience, actually balance-rr and balance-xor works just fine without any further switch configuration on any other switches than Cisco (OTOH I always have bizarre problems with Cisco switches...). balance-rr performance is OK with 2 ports, but sucks with more; CIFS doesn't play well with balance-rr, either. BUT balance-rr is the only mode were ...


3

He is partially right: Using LACP can increase availability due to it's automatic nature e.g. regarding dead links, but basically it's a protocol to control the bonding, regardless if it's used for performance enhancement or better resilience. So, your idea should work if all participating devices support it. ...


3

Is the link currently saturated? If not, then configuring Etherchannel is not likely to do anything for you. Link aggregation != better network performance (throughput) if the current link isn't saturated.


3

Well yes, setting up two NICs in a team only gives you failover redundancy, if it's done using an LACP-capable stack then etherchanneling on the switch allows you to, theoretically at least, double your bandwidth.


3

For the user facing network, STP is fine. Yes, you will have a small interruption as a new tree is computed. However, the convergence time is lower that the TCP timeout so the interruption should be effectively unnoticeable. Only for extremely time sensitive application such as VOIP would you run in to problems, and even those can be mitigated. For your ...


3

First question: What are the measured metrics used to determine your utilization (ie: what's "high" and "clogged"? I'm assuming this was done on the server only as you can't pull stats from unmanaged switches. Second: I would stay away from multiple nics on different subnets on the server. I would stay far away from the "introduce some small switches" ...


2

You should be using bonding mode - 4.


2

If your switch sees the true L3 destination, it can hash on that. Basically if you've got 2 links, think link 1 is for odd numbered destinations, link 2 is for even numbered destinations. I don't think they ever use the next-hop IP unless configured to do so, but that's pretty much the same as using the MAC address of the target. The problem you're going ...


2

4nics aggregation (2of4 active + auto failover): Linux: eth0-3 -> bond0, bonding mode=4 Switches (Juniper for example): port 1 + port 2 -> ae1 set interfaces ae1 aggregated-ether-options lacp passive set interfaces ae1 (...vlan assignments here...) passive on switch side ensures that it's Linux who'll device the partner to talk to. If you need PXE boot on ...


2

very suprisingly, a few days ago our testing showed that xmit_hash_policy=layer3+4 will not have any effect between two directly connected linux servers, all traffic will use one port. both run xen with 1 bridge that has the bonding device as a member. most Obviously, the bridge could cause the problem, just that it does not make sense AT ALL considering ...


2

I do know that a lot of new features are no longer supported on standard switches, such as the network rollback in case of misconfiguration, and network health checking. You can now save and restore your dVS separately, which I think was a big problem for people (which is why some would have recommended avoiding dVS entirely. I guess there are three reasons ...



Only top voted, non community-wiki answers of a minimum length are eligible