Does EIGRP loadbalancing affects the forwarding for specific destination address?

Hi,

Seems like I'm having this weird issue on my network that a specific IP block under switch1 vrf cust instance experiencing some delay or issue when accessing/connecting to a public proxy server and sometimes this cause delay when browsing using the said proxy server.

Updated Diagram: https://ibb.co/XFV0JMT

To make the story short, Did some isolation, check every segment interface etc.

From the above diagram you can see from the left part our current setup and core switch1 doesn't have any eigrp neighborship between R2(internet router) and we just use the Port-channel eigrp between HC2 using SVI 700 & 701.

​

Using this setup(left) , Client is experiencing slowness when accessing website using public proxy. Here's the routing from SW1 to proxy server.

SW1#sh ip route 123.1.1.1 Routing entry for 123.1.1..0/23 Known via "eigrp 1", distance 170, metric 2563072 Tag 100, type external Redistributing via eigrp 1 Last update from 3.3.3.2 on Vlan701, 00:03:07 ago Routing Descriptor Blocks: 3.3.3.2, from 3.3.3.2, 00:03:07 ago, via Vlan701 Route metric is 2563072, traffic share count is 1 Total delay is 120 microseconds, minimum bandwidth is 1000 Kbit Reliability 255/255, minimum MTU 1500 bytes Loading 1/255, Hops 2 Route tag 100 * 3.3.3.6, from 3.3.3.6, 00:03:07 ago, via Vlan700 Route metric is 2563072, traffic share count is 1 Total delay is 120 microseconds, minimum bandwidth is 1000 Kbit Reliability 255/255, minimum MTU 1500 bytes Loading 1/255, Hops 2 <---- Route tag 100 

Technically the routing looks good and since this is a slowness issue then it means Client can still connect to the public proxy to access websites. So EIGRP is doing loadbalancing between SW1 -> SW2 (SVI 700/701 eigrp adjacency).

​

I tried forming EIGRP adjacency between SW1 and R2(internet router), note that this is still via the same port-channel link between SW1 & SW2.

Now, Shows better result from client side (no delay when accessing website using public web proxy).

SW1#sh ip route 123.1.1.1 Routing entry for 123.1.1.0/23 Known via "eigrp 1", distance 170, metric 2562816 Tag 100, type external Redistributing via eigrp 1 Last update from 172.1.5.5 on Vlan999, 00:22:04 ago Routing Descriptor Blocks: * 172.1.5.5, from 172.1.5.5, 00:22:04 ago, via Vlan999 <- R2 INTERNET ROUTER Route metric is 2562816, traffic share count is 1 Total delay is 110 microseconds, minimum bandwidth is 1000 Kbit Reliability 255/255, minimum MTU 1500 bytes Loading 1/255, Hops 1 <---- Route tag 100 

Seems like using the EIGRP between HC1-HC2 SVI 700/701 somehow affects the forwarding towards to the proxy server, The thing here is why only specific destination is affected? The difference is the HOP from previous we are seeing 2 hops and now we are just 1 hop away to R2(internet router) because of EIGRP neighborship between SW1 and R2.

TRACE COMPARISON: W/ ISSUE SW1#trace vrf CUST 123.1.1.1 1 3.3.3.2 3 msec <---- SW1-SW2 EIGRP USING SVI 700/701 3.3.3.6 3 msec 3.3.3.2 4 msec 2 172.1.5.5 2 msec 2 msec 2 msec <---- R2 (INTERNET ROUTER) 3 * * * W/OUT ISSUE #trace vrf CUST 123.1.1.1 1 172.1.5.5 2 msec 2 msec 1 msec <---- R2 (INTERNET ROUTER) 2 * * * 3 * * * 

Note: technically SW1 is using the same physical link to forward the traffic between eigrp neighbor via Port-channel (refer to the diagram)

​

Question:

1. Does EIGRP load-balancing mechanism affects the forwarding to public proxy server? Note that other application works and it is using the same path.
2. Does the given next-hop affects the forwarding and why? We can see that from previous setup that route exist toward to public proxy(123.1.1.1) and the difference now is just before we are receiving the 123.1.1.0 public proxy block from HC2 (SVI 700/701) and now only R2 is selected as best path.
3. Please see the comparison between trace route, we can see that both trace able to forward the traffic to R2.

​

Seeking your technical inputs about this behavior?