Load Balance, Bundling ports

[mostafasawy86]

What is the difference between Load balance, and bundling ports in switch??

also, I don't understand if I have 4 links are bundled together, when the switch uses 1 link with hash algorithm, and when uses the all links together (in that case, is that Etherchannel Tech?)

I don't get it, please help me with CCNP Switch.

[Reza Sharifi]

Etherchannel provides load balancing based on source/destination IP, MAC, etc... The default for some switches (i.e. 6500 series) is source/destination IP.  So, if you have only one host on each side of an Etherchannle, since the source IP is only one, you will see the traffic will use only one of the links.  Say, you have 2 hosts on each side of the Etherchannedl, now one host uses one link and the other one uses a different link in the bundle.

HTH

[mostafasawy86]

Also, how to configure the switch in the following cases:

1-if I have 4 links, and I want to load balance them so the packets transmitted to distination with selected link.

2-if I have 4 links, and I want to bundle them as 1 link, so the packets transmitted to destination using all links.

[Reza Sharifi · Accepted solution]

1-if I have 4 links, and I want to load balance them so the packets transmitted to distination with selected link.

You can't configure that.  It follows a sepcific hasining algorithem and uses that. If you have 4 links host one uses the first link, the second host the second link, and so on.

2-if I have 4 links, and I want to bundle them as 1 link, so the packets transmitted to destination using all links.

A flow can only use one of the links and not all four at the same time

HTH

[mostafasawy86]

so, in other words, it only uses 1 link at a time, and I can't use 2 parallel link at a time?

i.e, if I'm transmitting data, and during the transmission, I disconnected that link, the transmission won't be complete, so I need to re-transmit it again so it can use the second line?

[Reza Sharifi]

Correct, you loose the flow that is already in the wire.  Now, if everything is configured correctly and if one link fails within a second or so, it would fail over to the other link. If you run a continuous ping, and pull a physical link out, you should not loose more than one to ping packets before it fails to the second link.

HTH

[mostafasawy86]

so that is different from equal load sharing, and unequal load sharing that is done by routers with EIGRP or OSPF?

[mostafasawy86]

also, the same concept goes for LACP and PAgP?

[Reza Sharifi]

yes, for OSPF is different

also, the same concept goes for LACP and PAgP?

Yes, whether LACP or PAGP, they both work the same way.

[mostafasawy86]

for OSPF, if a router has 2 route to a destination, either it load shars per destination or per packet.

either using a link for dedicated destination, or use both links for distributing the packets?

I really appreciate your help, you really helped me to get what I miss.

also, you helped me to get some rest so early

[Reza Sharifi]

Good luck with your exam:

Regarding OSPF load balancing:

Per-Destination and Per-Packet Load Balancing

You can set load-balancing to work per-destination or per-packet.  Per-destination load balancing means the router distributes the packets  based on the destination address. Given two paths to the same network,  all packets for destination1 on that network go over the first path, all  packets for destination2 on that network go over the second path, and  so on. This preserves packet order, with potential unequal usage of the  links. If one host receives the majority of the traffic all packets use  one link, which leaves bandwidth on other links unused. A larger number  of destination addresses leads to more equally used links. To achieve  more equally used links use IOS software to build a route-cache entry  for every destination address, instead of every destination network, as  is the case when only a single path exists. Therefore traffic for  different hosts on the same destination network can use different paths.  The downside of this approach is that for core backbone routers  carrying traffic for thousands of destination hosts, memory and  processing requirements for maintaining the cache become very demanding.

Here is the link:

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094820.shtml#perper

HTH