High latency in the Colombian region

Questions / Help
1

A few days ago I noticed that there is a new region of Colombia, Bogota, a couple of days ago it said that there were no disks available, today I tried again and everything seems to work, but the latency is higher than expected.

I’m in Colombia and although the fly launch suggests Bogota to me, when I do some tests I always have higher latency, compared to, for example, Miami when I make an HTTPS request to my service.

Using the ping command I do see a slight improvement (as expected):
Miami: 62.325 ms
Bogota: 56.833 ms

But going into the service from the browser, in the exact same Phoenix app, at the same healthcheck endpoint, the “waiting for server response” is:
Miami: 62ms
Bogota: 124ms

And in the metrics graph I got this, look at the average in the bottom:
Miami:

firefox_HLbEj9ver4
firefox_HLbEj9ver41138×395 24.4 KB

Bogota:

firefox_ClkttQ54mj
firefox_ClkttQ54mj1183×397 19.5 KB

So I can only think that the servers there are lower in performance, or internally in the DC there is some network latency, or the region isn’t production ready yet?

2

Is this app making requests to a database or other service that lives in a single region? That could explain the latency.

3

It was a simple phoenix app, that indeed had a connection to a DB, the endpoint only had one query (“select 1”), as means of health check. Anyways the DB was in the same region.

I just did fly launch, in each region, also creating the DB in the same command. Unless fly launch created the DB in another region, I didn’t check that, but it is assumed that if you select a region, both app and DB are created in it.

No other external services on the endpoints that I was testing.

PD: could be great if you can add Bogota to https://liveview-counter.fly.dev/, as it seems to display real latency. So, I can test there too.

1 Like