Why does OSMF Budget €25,000 on Amazon

Right now I understand that we pay €25000 yearly on storing primarily CDN logs and backups? Is that correct? This can’t be right.

From everything I’ve read in this threat so far you’re wrong. Can you step back, collect your thoughts and make a new post with your finding and/or questions in a couple of days? On github you edited one comment 8 times, it’s hard to follow. And as another user already wrote it reads like you’re in attack mode (knowingly or not), every other sentence reads like another assusation that other people hide something.

7 Likes

@NorthCrab Please let Firefishy get back from holiday and give you a full answer rather than keeping posting here about some links you might have found on the OSMF wiki.

For some more context, I edited this github comment (edits are publicly accessible) because I initially couldn’t find more information about the bi-weekly OWG meetings. After discovering this page: All pages with prefix - OpenStreetMap Foundation, I decided to retract my statement. This page is quite hidden on the OWG website: https://operations.osmfoundation.org.

We don’t yet use AWS for serving planet data, the ticket is still open for a reason :wink: And when we do it will be $0/month due to AWS sponsoring the cost under the AWS public data program. We have a dedicated AWS account setup with delegated billing to AWS already setup.

Maybe you are right, but it’s hard to remain calm when you find such conflicting statements.

  1. AWS has been free for 6 months now: Why does OSMF Budget €25,000 on Amazon - #4 by Firefishy
  2. Wait, actually it WILL be free: Operations/Minutes/2023-01-12 - OpenStreetMap Foundation, when we finish Move planet hosting to S3 · Issue #678 · openstreetmap/operations · GitHub first (quote: “Get planet tasks done and then ask for AWS sponsorship”)

Funny thing, I never asked whether it’s free or not. I accessed public information and found multiple concners with it, which I decided to bring up. I don’t know what the argument “hey buy it’s free now” really does here. It doesn’t answer anything.

Here is a screenshot from the AWS which actually shows 25,000€ spending estimate for 2023. The core issue I have, is why? Why so much? What do we store there? So far I only got this:

  • Primarily CDN logs
  • Primarily backups
  • Secondary GPX traces
  • Secondary profile pictures

Those should not be a 25,000€ yearly spending. That’s what I am most concerned about. This is not right. That’s what I want to bring attention to.

Please don’t change my wording. I said:

It’s quite obvious that I don’t have access to private information of others.

So, you have financial accounting and management accounting. In management accounting is just a way to allocate funds into different groups. To make an estimation of what might be spend. you want to make sure that your budget is always higher than what you might spend on that group. In this case you have a group where the spending is uncertain it is better to overestimate the spending (conservative accounting)

By estimating conservatively, you make sure you do not overspend, and when you make up the balance sheet (financial accounting) it looks a lot better to be able to write some money to the right side. Instead of having to cut into reserves.

1 Like

The core issue is that OWG don’t release the “full story” and that’s what concerns me the most.

That’s exactly what I mean. The topic wants to know more about OWG AWS costs which are private/hidden.

As per, Estimate cloud costs for 2023 · Issue #788 · openstreetmap/operations · GitHub, those are/were real spendings and the estimation for 2023 was quite on point. The difference between estimation and budget from the attached image, is quite negligable.

Wait until you see how much we spend with equinix (data centres), he.net (ISP) and on physical server hardware. :wink:

I don’t think you quite get the scale of the project.

I can breakdown our cloud usage / expenses / sponsorship once I back on Wednesday and have access to the real data.

5 Likes

Why do you deflect my questions? I don’t ask about other servers. I ask specifically about the AWS 25,000€ yearly spending. You have deflected my questions numerous times now which does not ease the situation.

Your comments are off-putting, let the person be on holiday and calm down. You’re also deflecting and ignoring the "I can breakdown our cloud usage / expenses / sponsorship once I back on Wednesday " reply you got.

11 Likes

I won’t be mad if I don’t get an answer for a week or so (I don’t expect I will get an answer today). I am just replying to new posts that come in. Should I stop replying then?

I have nothing to say on that yet, I am patiently waiting for Wednesday.

Hi @NorthCrab, I encourage you to read the guidelines for interacting on the Community forum. As others have said, it’s entirely off-putting to “come in swinging”. If you assume good faith to start with, people will be much more likely to participate in a healthy discussion.

As it is, your messages make people feel defensive and annoyed. Please change your tone of voice before continuing with this thread.

21 Likes

Hey @iandees, if you quote me specific texts that break the guidelines, that would help!

I am typing on my phone, trying to best answer your questions without being able to login into AWS console.

We have a mountain of backups (secondary + archival), logs (for processing) and live user/gpx/images files used for website. At very least 70TB+. As discussed I can give you real numbers later in the week. Physical hardware might be cheaper, but running it would suck up our limited time, us being an extremely small team.

We are using credits from AWS. We think we will get free credits again but have a budget estimate if we didn’t and spent like we have (on credits). In reality we would find some way to cut back if we did not get credits. Since Jan 2023 our total real AWS expenditure has been $0 due to usage of credits. Azure is around $30/month for translation API used by this forum.

13 Likes

So 70TB of which:

  • 4TB are logs (for processing)
  • 65TB are backups
  • 1TB others, secondary things

Is that correct? Why are backups taking so much space? Is some deduplication logic applied or are all backups completely separate of each other? Those backups better be good for a 25,000€/y solution :smiley:.

Have you ever considered non-AWS solutions? AWS is in general very expensive. Backblaze is a reputable backup company with years of experience, providing similar API interface, costing just 1/5th of AWS: Cloud Backup Pricing for Veeam, NAS, Servers, and Workstations. Such a simple switch would reduce 25,000€ to a more reasonable ~5,000€ range. A 20% save in total yearly OWG spendings.