CCNP learning notes for the round-robin Load Balancing Based on Network Replication

 

Non-equivalent Server Load balancer based on Principle : Successor FD * variance> feasible successor FDTopology

650) this. width = 650; "src =" http://img1.51cto.com/attachment/201202/212008309.jpg "border =" 0 "alt =" "/>

After completing the configuration on the three routers, the basic configurations are not listed here, and then the lookback 0 with the IP address 2.2.2.2 is declared on R2.

Purpose

To go to 2.2.2.2 this network segment to R3, you must first check the route table of R3 by creating the two unequal Load Balancing rules R3> R2 and R3> R1> R2.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P02561-1.jpg "border =" 0 "alt =" "/>

On R3, view the detailed neighbor table sh ip address (all ).

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P0La-2.jpg "border =" 0 "alt =" "/>

We can see that there are two routes in the 2.2.2.0 CIDR block, while S1/0 and E1/0 respectively, FD passing through E1/0 is much larger than FD passing through S1/0, and AD of S0/1 is much larger than FD of E1/0, so no to become a feasible successor, but in load balancing that is not equivalent, feasible successor must be available, here, I am modifying the E1/0 latency of R3 to manually adjust S0/1 to feasible successor. Here is a formula FD metric = 256*10 ^ 7/BW + DLY. /10) view default latency of E1/0

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P045P-3.jpg "border =" 0 "alt =" "/>

The estimated latency is set to 8000. Enter the interface R3 (config-if) # delay 8000 to get the detailed neighbor table of R3

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P03056-4.jpg "border =" 0 "alt =" "/>

Here, S0/1's AD 2297856 is less than the FD 2432000 of E1/0, so S0/1 becomes the feasible successor. In actual projects, the latency will not increase .. That's what SB Does. In actual projects, SLB is not implemented through the S port and E port -- HereOnly for demonstration) and then calculate the variance value through the principle of non-equivalent load balancing Principle : Successor FD * variance> feasible successor FD Here 2432000 *The variance value must be greater than 2809856. Therefore, set this parameter to 2. Enter the OSPF process R3 (config-router) of R3 # variance 2 and check the route table of R3.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P04327-5.jpg "border =" 0 "alt =" "/>

Go2.2.2.0The specified CIDR block already exists.2Route entry

InPingPass2.2.2.2After the address, View2.2.2.0 255.255.255.0Detailed information about this route

 

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131227/021P02235-6.jpg "border =" 0 "alt =" "/>

We can see that the 23.1.1.2-based data packet passed through 15 data packets 13.1.1.1 passed through 13 Load Balancing because they are not equivalent, so there is a balance algorithm to allocate the data packet path, while FD is smaller, FD is smaller. How is this balancing algorithm calculated .... Don't bother to worry about him... Back pain. Go to rest...

 

This article from the "lustlost-lost in desire" blog, please be sure to keep this source http://lustlost.blog.51cto.com/2600869/770734