I was looking for these to show off to a friend (I said the funding drive was mostly for boxes and bandwidth, they were curious as to just how much we actually served) and couldn't find them. I thought they were on http://ganglia.wikimedia.org/ but that appears to be CPU and memory. Are the network and bandwidth charts still up anywhere?
- d.
On 18/12/06, David Gerard dgerard@gmail.com wrote:
I was looking for these to show off to a friend (I said the funding drive was mostly for boxes and bandwidth, they were curious as to just how much we actually served) and couldn't find them. I thought they were on http://ganglia.wikimedia.org/ but that appears to be CPU and memory. Are the network and bandwidth charts still up anywhere?
Mark appears to have some that are updated at http://www.nedworks.org/~mark/reqstats, e.g. http://www.nedworks.org/~mark/reqstats/reqstats-hourly.png - is this what you're after?
Rob Church
On 18/12/06, Rob Church robchur@gmail.com wrote:
On 18/12/06, David Gerard dgerard@gmail.com wrote:
I was looking for these to show off to a friend (I said the funding drive was mostly for boxes and bandwidth, they were curious as to just how much we actually served) and couldn't find them. I thought they were on http://ganglia.wikimedia.org/ but that appears to be CPU and memory. Are the network and bandwidth charts still up anywhere?
Mark appears to have some that are updated at http://www.nedworks.org/~mark/reqstats, e.g. http://www.nedworks.org/~mark/reqstats/reqstats-hourly.png - is this what you're after?
That's the sort of thing :-) I just remember from a year or two ago a scary set of Ganglia-generated graphs with a worrying upward trend ...
- d.
On 18/12/06, David Gerard dgerard@gmail.com wrote:
That's the sort of thing :-) I just remember from a year or two ago a scary set of Ganglia-generated graphs with a worrying upward trend ...
I remember them as well; there were some on Kohl, I think, in home directories. It's possible those got lost/disabled somehow when we had that whole NFS mount business?
Rob Church
On 18/12/06, Rob Church robchur@gmail.com wrote:
On 18/12/06, David Gerard dgerard@gmail.com wrote:
That's the sort of thing :-) I just remember from a year or two ago a scary set of Ganglia-generated graphs with a worrying upward trend ...
I remember them as well; there were some on Kohl, I think, in home directories. It's possible those got lost/disabled somehow when we had that whole NFS mount business?
Sounds familiar. I ask because I put a begging notice on my livejournal:
http://reddragdiva.livejournal.com/375133.html
- and the geeks promptly asked for detailed info. As they would be expected to.
(And I realise Alexa numbers are less than widely respected, but still consider that they give us some idea of scale, i.e. we're currently in the range of 1/5 to 1/10 Google's page views. JESUS CHRIST IT'S A LION GET IN THE CAR)
- d.
On 12/18/06, David Gerard dgerard@gmail.com wrote: [snip]
(And I realise Alexa numbers are less than widely respected, but still consider that they give us some idea of scale, i.e. we're currently in the range of 1/5 to 1/10 Google's page views. JESUS CHRIST IT'S A LION GET IN THE CAR)
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
It is indisputable, however, that we are one of the more visited sites around.
On 18/12/06, Gregory Maxwell gmaxwell@gmail.com wrote:
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
When it comes down to convincing people to give us cash, what's a little unreliable data among people who know better?
Rob Church
On 18/12/06, Rob Church robchur@gmail.com wrote:
On 18/12/06, Gregory Maxwell gmaxwell@gmail.com wrote:
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
When it comes down to convincing people to give us cash, what's a little unreliable data among people who know better?
They'll be reading this thread and want the real numbers? ;-)
- d.
On 18/12/06, Rob Church robchur@gmail.com wrote:
On 18/12/06, Gregory Maxwell gmaxwell@gmail.com wrote:
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
When it comes down to convincing people to give us cash, what's a little unreliable data among people who know better?
They'll be reading this thread and want the real numbers? ;-)
- d.
On 12/18/06, David Gerard dgerard@gmail.com wrote:
That's a pity - 70,000 users is really quite a high-resolution sample [*]; it's a pity their sample selection is evidently crappy.
Three cheers for the self-selecting sample.
On 12/18/06, David Gerard dgerard@gmail.com wrote:
They'll be reading this thread and want the real numbers? ;-)
Well, the link has been given... and the same data has been previously provided to other donors, it's just that it never made it onto the pages.
The graphs themselves are a bit hard to read because the daily volatility swamps the short term growth... If only I there were stocks which I could trade that followed our traffic patterns.
Keep in mind that roughly 2/3 of the overall requests are images, so a correction must be applied to convert req/s to approximate page views/src.
On 12/18/06, Gregory Maxwell gmaxwell@gmail.com wrote:
On 12/18/06, David Gerard dgerard@gmail.com wrote: [snip]
(And I realise Alexa numbers are less than widely respected, but still consider that they give us some idea of scale, i.e. we're currently in the range of 1/5 to 1/10 Google's page views. JESUS CHRIST IT'S A LION GET IN THE CAR)
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
The key is *somewhat*. There are probably very few sites ranked under, say, 30-50 (or at least under 100) that actually get more traffic than Wikipedia. Alexa is acceptable for very rough ballparks, although how rough exactly would be interesting to statistically analyze.
On 18/12/06, Gregory Maxwell gmaxwell@gmail.com wrote:
On 12/18/06, David Gerard dgerard@gmail.com wrote:
(And I realise Alexa numbers are less than widely respected, but still consider that they give us some idea of scale, i.e. we're currently in the range of 1/5 to 1/10 Google's page views. JESUS CHRIST IT'S A LION GET IN THE CAR)
You're giving too much credibility to the alexa data. There are sites which are ranked somewhat lower than us which I know for a fact are processing more req/s.
That's a pity - 70,000 users is really quite a high-resolution sample [*]; it's a pity their sample selection is evidently crappy.
It is indisputable, however, that we are one of the more visited sites around.
The traffic stats yearly graph makes that clear. 1 gigabit/second sustained, doubling in the last six months ... if we could just slow it down to Moore's Law rates ...
[*] http://en.wikipedia.org/wiki/Margin_of_error - in a result which many statistics students find surprising, the margin of error depends *only* on the number of samples taken, not on the size of the sampled population. Most political polls, for example, sample about 2000 people; sampling 70k is a *lot* of people.
- d.
http://www.nedworks.org/~mark/reqstats/ is where Mark Bergsma has our traffic information. Pay attention to the monthly and yearly graphs; the trendline is unmistakeable.
David Gerard wrote:
I was looking for these to show off to a friend (I said the funding drive was mostly for boxes and bandwidth, they were curious as to just how much we actually served) and couldn't find them. I thought they were on http://ganglia.wikimedia.org/ but that appears to be CPU and memory. Are the network and bandwidth charts still up anywhere?
- d.
Wikitech-l mailing list Wikitech-l@wikimedia.org http://mail.wikipedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org