High internal bandwidth usage

I started running a small personal app on Fly early in December. The bandwidth consumption seems really high - I used 834GB in December, which is big chunk of my relatively small bill.

I think a lot of my consumption is internal to the main (and only) region. The app makes a lot of requests to the Twitter API, but then writes the results to the Postgres DB in an upsert manner, resulting in a lot of internal traffic, which I haven’t had to account for previously.

I’d really like to stay on Fly, but this has taken me a bit by surprise!

Is there any easy way to confirm how much of my traffic is internal vs external?

And do you have any plans to split internal/external traffic pricing?

1 Like

That is a lot! We’ll look into your account this week. There’s no great way to see the breakdown of bandwidth, but we should be making bandwidth between your app and postgres as free as possible.

Will you please bump this post if we haven’t gotten back to you by Thurs?

Of course, thanks!

Hi there,

I’m actually having a similar issue. I’ve launched a small project (with a staging + prod environments) and i was also surprised to see that i’m getting a bill of more than 100$ which is linked mostly to outbound bandwidth. From the look of it, it’s linked to both of my PG DB. What i don’t understand is that i actually don’t have much outbound traffic, i believe it’s most internal transactions between my app and my DB.
I initially thought this would be because of logging my app’s logs onto a 3rd party services. I disabled it and i’m still getting the same bill.
In December i had 4TB from my 2 DBs (staging and prod) :flushed:

Am i doing something wrong?


Thanks for the bump!

We’re going to account for bandwidth from Postgres differently starting this month. I’ll refund the bandwidth on both of your invoices for December. We don’t actually know how much was Postgres vs App outbound, but refunds are easy. :wink:

Thanks! Do you have any plans to treat bandwidth inside a region differently, more generally? It doesn’t feel right to me that moving stuff to/from Redis in the same region, or between same region BEAM nodes, should be billed the same as an end user.

Yes, but the underlying costs are incredibly complicated. We’re going to try not charging for data transfer unless an app has public IPs for a few months. If we can survive doing things that way, it’ll make everyone’s life easier. :smiley:

1 Like

That’s great to hear! It also happened to me on November as well :sweat_smile: :blush:

Thanks for the support!

@kurt I too am seeing high bandwidth usage for an app that does not have much traffic. Can you please look into my org to see if I’m doing something wrong?

1 Like