Big difference in network latency of 1st gen vs 2nd gen

[Son Nguyen]

I have 4 SQL instances in "us-central1-b" (same region, same zone), two 1st gen (173.194.xx.xx) and two 2nd gen (104.154.xx.xx, 130.211.xx.xx)

Network latency from my app server (Dallas, TX) (similar results from other locations)

173.194.xx.xx => 7ms (1st gen)

104.154.xx.xx => 23ms (2nd gen)

130.211.xx.xx => 28ms (2nd gen)

Surprise! 1st gen has much better and faster network latency that 2nd gen. 

I've also tried to change the zone within us-central and different IPs but same result.

Anyone knows why there is such a big difference in latency (7ms vs. 23/28ms)?

Thanks

S

[Adam (Cloud Platform Support)]

Is your app server onsite or hosted? Can you share details as to how you are measuring latency?