Out of curiousity...I wonder if anyone has an idea of how many centos installations are out there?
On Wed, 2005-06-01 at 21:23 -0400, Todd Blake wrote:
Out of curiousity...I wonder if anyone has an idea of how many centos installations are out there? _______________________________________________
169,788 distinct IP addresses have asked for updates since 04-Mar-2005 from mirror.centos.org
That doesn't include anyone who does their updates from one of the external public mirrors (current 45 listed).
Johnny Hughes schrieb:
On Wed, 2005-06-01 at 21:23 -0400, Todd Blake wrote:
Out of curiousity...I wonder if anyone has an idea of how many centos installations are out there? _______________________________________________
169,788 distinct IP addresses have asked for updates since 04-Mar-2005 from mirror.centos.org
That doesn't include anyone who does their updates from one of the external public mirrors (current 45 listed).
Hi,
I'm glad, nobody started to count directly, like "one" (me) :-) Apart from that I don't think, it's a good measure to count the IP addresses "since". A better measure might be to release _one_ update and to count, how many guys download that update within say two weeks, morover to ask the mirror guys to give away their access data for this update. The best way might be to implant an easter egg, which connects to a kind of a counter server, but this might lead to many discussions... cu - Michael
On 6/1/05, Michael Kress kress@hal.saar.de wrote:
Johnny Hughes schrieb:
169,788 distinct IP addresses have asked for updates since 04-Mar-2005 from mirror.centos.org
<snip>
Holy moly that's a lot.
<snip>
Apart from that I don't think, it's a good measure to count the IP addresses "since". A better measure might be to release _one_ update and to count, how many guys download that update within say two weeks, morover to ask the mirror guys to give away their access data for this update.
There's a couple of flaws with a "distinct IP" count:
1. Not all servers get updated (I know of 3 or 4 CentOS servers that probably haven't been updated in months. 2. One machine can get multiple IPs 3. Many machines can look line one IP
However, distinct IP does give us a rough estimate of installs. And even if it's off by 30% one way or another - it's a bigger number than I thought.
Greg
On Thu, 2005-06-02 at 12:26 -0600, Greg Knaddison wrote:
<snip>
There's a couple of flaws with a "distinct IP" count:
- Not all servers get updated (I know of 3 or 4 CentOS servers that
probably haven't been updated in months. 2. One machine can get multiple IPs 3. Many machines can look line one IP
Right ...
One big issue also is that the distinct IPs measured were only updates done to CentOS mirrors ... any updates done from external mirrors are not counted in any way ... so, that number is probably low by a large factor.
Just some other numbers to look at:
Bittorrent ISOs downloaded in 2 months: 32778 (that doesn't include any ISOs downloaded from external mirrors)
CentOS.org website traffic rating at alexa.com: 36,829 (that is better than several big name distros like Knoppix, Slackware, Xandros, Mepis, PCLinuxOnline, etc.)
Average traffic out of CentOS mirrors is 16.5 TB per Month.
On Thu, 2005-06-02 at 19:05 -0500, Johnny Hughes wrote:
On Thu, 2005-06-02 at 12:26 -0600, Greg Knaddison wrote:
<snip>
Just some other numbers to look at:
Bittorrent ISOs downloaded in 2 months: 32778 (that doesn't include any ISOs downloaded from external mirrors)
CentOS.org website traffic rating at alexa.com: 36,829 (that is better than several big name distros like Knoppix, Slackware, Xandros, Mepis, PCLinuxOnline, etc.)
Average traffic out of CentOS mirrors is 16.5 TB per Month.
(That is also only the CentOS.org mirrors ... not any traffic out of the close to 50 external public mirrors)
Johnny Hughes wrote:
One big issue also is that the distinct IPs measured were only updates done to CentOS mirrors ... any updates done from external mirrors are not counted in any way ... so, that number is probably low by a large factor.
I agree, I think, most people would chose a "more local" mirror in order to get faster updates. Michael