3 ms·

The author also failed to notice "peaks" from the original spotify blog post about events per second. So 700k requests is not average, it's peak. They don't hav

The author also failed to notice "peaks" from the original spotify blog post about events per second. So 700k requests is not average, it's peak. They don't have that every second.

stephenitis 11y ago
nice point. If you want to further refine you could probably estimate a large chunk of early morning US/EU time where traffic will be very light. maybe reduce the estimate by 1/3 or 1/4th to account for this.
dmytton 11y ago
This is a good point, but it could be somewhat mitigated by their expected growth of up to 2M per second. To quote the full line from their blog post: > Currently our production load peaks at around 700K events per second. To account for the future growth and possible disaster recovery scenarios, we settled on a test load of 2M events per second. In the absence of any other numbers, using the 700k number seems reasonable otherwise you would start to get into modelling without any real data to back up your assumptions. This at least gives you a maximum they would be paying under current production conditions.
- bkovacev 11y ago
  That's actually a reasonable response. From my first read the numbers seemed a bit inflated, so I dug into their blog post out of curiosity (I develop on GAE), so the comment was not ill intended. Now that you have pointed out they do seem a good estimate! Thanks for the great articles David, keep them coming! :)
- mkagenius 11y ago
  How about a gaussian distribution which peaks at 700k? Assuming max as the norm is not too right. (just to make the number look high)
  - EricBurnett 11y ago
    You could guess that, but with no real expectation that it's right either. E.g. for a global service I'd guess it's better viewed as a sum of three offset distributions, one per continent, but even then the individual distributions are unlikely to be gaussian. Having run global services myself I can tell you that the shape is usually no standard distribution. Peak/trough depends on the exact usage demographics, but the one I have on hand has a trough of ~1/2 peak, and an average of ~3/4.
    - smarinov 11y ago
      Just curious: how does the real distribution look like, from your experience? Do you have any graphs by any chance?
- ej_campbell 11y ago
  If this is the logic, then a better title would have been "How much will Spotify be paying Google Cloud in a few years, assuming 3x growth from today's numbers?" It's really not fair to tell the world that their costs are based on average of 700k rps, when that is their peak traffic through the system.

The author also failed to notice "peaks" from the original spotify blog post about events per second. So 700k requests is not average, it's peak. They don't have that every second.

stephenitis 11y ago

nice point. If you want to further refine you could probably estimate a large chunk of early morning US/EU time where traffic will be very light. maybe reduce the estimate by 1/3 or 1/4th to account for this.

dmytton 11y ago

This is a good point, but it could be somewhat mitigated by their expected growth of up to 2M per second. To quote the full line from their blog post: > Currently our production load peaks at around 700K events per second. To account for the future growth and possible disaster recovery scenarios, we settled on a test load of 2M events per second. In the absence of any other numbers, using the 700k number seems reasonable otherwise you would start to get into modelling without any real data to back up your assumptions. This at least gives you a maximum they would be paying under current production conditions.

bkovacev 11y ago
That's actually a reasonable response. From my first read the numbers seemed a bit inflated, so I dug into their blog post out of curiosity (I develop on GAE), so the comment was not ill intended. Now that you have pointed out they do seem a good estimate! Thanks for the great articles David, keep them coming! :)
mkagenius 11y ago
How about a gaussian distribution which peaks at 700k? Assuming max as the norm is not too right. (just to make the number look high)
- EricBurnett 11y ago
  You could guess that, but with no real expectation that it's right either. E.g. for a global service I'd guess it's better viewed as a sum of three offset distributions, one per continent, but even then the individual distributions are unlikely to be gaussian. Having run global services myself I can tell you that the shape is usually no standard distribution. Peak/trough depends on the exact usage demographics, but the one I have on hand has a trough of ~1/2 peak, and an average of ~3/4.
  - smarinov 11y ago
    Just curious: how does the real distribution look like, from your experience? Do you have any graphs by any chance?
ej_campbell 11y ago
If this is the logic, then a better title would have been "How much will Spotify be paying Google Cloud in a few years, assuming 3x growth from today's numbers?" It's really not fair to tell the world that their costs are based on average of 700k rps, when that is their peak traffic through the system.