Arduino about to hit "10.000.000 downloads" only it's page hits

[alexanderbrevig]

So, while I've been paying attention this last day - according to their counter - they've served ~ 100 10 Terabytes of data.
Seems unlikely?

Agreed.

So, do this:
Refresh this a couple of times: https://www.arduino.cc/en/Main/Software
And I will guarantee you that this counter https://www.arduino.cc/download_handler.php?getCount will increase at least as much.

Don't know why it bothers me, but it does. It's a straight out lie to call it downloads when it's probably mostly robots visiting the download page. 

EDIT: made a mistake, it's 10 tera, not a 100 (forgot to check the math  ) - still way too much for a day worth of downloads

[wraper]

I don't see any causation between reloading that page and counter increase. I can just wait the same amount of time and refresh the counter, counter increase will be about the same. Or I can refresh download page 10 times but counter will increase only by half of that, just because of the time delay I spent on refreshing.

100 Terabytes of data

Not so much, frankly.

[wraper]

When thinking about 10 million downloads. What is population of earth, 7 billion. Then it gets to 1 download per 700 people. Considering that one person can download multiple times and the amount of cheap Chinese arduino clones in the wild, this figure does not seem strange for me.

[alexanderbrevig]

100 Terabyte bandwidth for one day don't seem much to you?
I'll remember to hire you if I need a godaddy set up

[wraper]

100 Terabyte bandwidth for one day don't seem much to you?
I'll remember to hire you if I need a godaddy set up

Where it or your post says it's a daily counter? And the speed of it's rise does not suggest it can get 10 million counts even in month.

[alexanderbrevig]

[...] this last day [...] they've served ~ 100 10 Terabytes of data.

I apologize for the hard to parse sentence. I can see how it was not clear.
It was a simple matter of multiplying the difference in count from time A to time B with the average size of the downloads.

[wraper]

Ok, so it's

SINCE MARCH 2015, THE ARDUINO IDE HAS BEEN DOWNLOADED 9,921,887 TIMES. (IMPRESSIVE!)

Considering 84 MB size, it's 833.5 terabytes during those 1.5 years result in about 1.5 TB per day.

[EEVblog]

10 million sounds right.
Many people would have downloaded it several times over the years.
Countless clones etc.

[alexanderbrevig]

Yes, only the download was much bigger before they added the board manager and things like that.
It's an unnatural increase in counter speed for this final push toward 10mill.

I just found it interesting, for me when I test there still is a strong correlation between refreshes and the count.
When it was braught to my attention I made a script and this happened:


I will bet a lot of money most of those hits are from my script. Seems to average around 18k per day, so 81k/16h is definitly an anomaly.

It's only interesting to me, don't mean to come off like this is important or a huge conspiracy.

[LabSpokane]

10 million sounds completely plausible. And if you don't like Arduino.cc's download count, why don't you demand a refund?

Oh, that's right, you didn't pay for the software in the first place. 

[rrinker]

 I've personally downloaded it at least 5 times across multiple computers - call it spoiled by fast internet. In the old days, I would have downloaded it once and saved it on my server, and installed it on other computers from that one download. Now, I just download things I need and only save some of the really big things that would take significant time to download. So I can easily believe that the Arduino IDE has been downloaded 10 million times.

[dr.diesel]

And that counter doesn't include all of the installations from Linux repos.

[amspire]

It sounds like a lot of data, but they are hosted on Amazon and the cost is less then the cost of hiring one employee. It is all very manageable.

[Seekonk]

Could be the noobs have to download it ten times before they figure out how to make it  work.

[Nerull]

This seems...strangely obsessive. It sounds like you've got some issues.

Arduino is very popular. 10 million downloads is fairly plausible.

For comparison, they've sold over 10 million Raspberry Pi's, for which the market is not flooded with cheap clones.

Websites with millions of hits per day are now so bloated that the webpage itself is several megabytes, before any media. Can you imagine how much bandwidth that is?

[mariush]

Ok, so it's

SINCE MARCH 2015, THE ARDUINO IDE HAS BEEN DOWNLOADED 9,921,887 TIMES. (IMPRESSIVE!)

Considering 84 MB size, it's 833.5 terabytes during those 1.5 years result in about 1.5 TB per day.

It sounds like a lot of data, but they are hosted on Amazon and the cost is less then the cost of hiring one employee. It is all very manageable.

You can rent dedicated servers with 1 gbps unmetered internet connections for about $100-130 euro a month, and that would basically give you about 300 TB of bandwidth each month, or about 30 TB a day. Here's an example: https://www.leaseweb.com/bare-metal-server/configure/24570#contract-container

In this particular case, a CDN may be a good solution to use for small periods of time right after a new version of the software is released when you're going to have a burst of downloads for a few days, but afterwards it may make more sense to rely on your own servers.

For example, a CDN would charge something like 5 US cents per GB of delivered content, so for 1.5 TB a day, you'd pay $90-100 a day with CDNs, or around $3000 a month.  You could literally rent half a rack with a 1gbps unmetered connection for less than $2000 .. put a bunch of servers there to serve your files directly from ram or from a SSD and you're done (though you'd make your service vulnerable to ddos attacks).

Anyway, it's a shame they don't offer torrents for the latest releases. They could make a single torrent for the latest release, with each operating system and 32/64 bit versions in separate folders, so users could select and download all or just one

[amspire]

They are on Amazon and I think without CDN, since I got the same IP address from Australia and the US. They are also enforcing HTTPS so they cannot take advantage of local ISPs caches.

But as I said, the Amazon traffic cost is less then the wage of one employee so it is manageable.

The download quantity is probably lower then I would expect seeing that since March 2015, there have been 11 versions released, and for each version, there are separate downloads for Windows, 32 bit Linux, 64 bit Linux and MAC. I am not a big Arduino user, but I have probably downloaded 3 or 4 times over the period. A keen user may have downloaded all 11 releases for multiple platforms.

It is definitely not the case that there have been 10 million new users over the year. Perhaps 1 million new users. Basically every time someone buys a cheap $5 Arduino clone on eBay or Aliexpress, they are going to download the latest development kit fron Arduino.cc.

[alexanderbrevig]

Yeeeah, I had forgotten about this thread.
Here's my apology for starting a negative and speculative thread.

I've probably used Arduino as much as any here and have a lot of libraries online. There's also the fact I've been on the wrong side of the untold history of Arduino, as a core memeber of Wiring.
You also used to be able to get to me at alex@arduino.org (I knew about the merger at the time so .cc / .org had nothing to do with this).

This whole thing was just me noticing the counter behave in a weird way, and I found it interesting. I'm sorry I did not stop to think if it really was for anyone else, and for marinating it in my personal frustration with the company.
I love and promote the idea and the framework behind Arduino, whenever it's appropriate (and admittedly maybe sometimes when it's not).

Correlation doesn't imply causation, I know this - and should not have made this thread.