Hmmm. What about one large telco (blue team) taking five days to acknowledge a problem with their international links and cache farm? Or another large telco (red team) taking a few months to acknowledge radio problems? Their reason was "no one logged faults". Of course not - users called, CSRs asked to power cycle their fancy iPhones, it works after that - until it happens again a few hours later. Since it "worked" the CSRs didn't log faults. I think 72 hours is a good sided window... Mauricio Freitas http://www.geekzone.co.nz http://www.geekzone.co.nz/freitasm http://www.twitter.com/freitasm -----Original Message----- From: Drew Broadley [mailto:drew(a)broadley.org.nz] Sent: Tuesday, 7 December 2010 10:57 a.m. To: Mauricio Freitas Cc: Nathan Ward; nznog(a)list.waikato.ac.nz Subject: Re: [nznog] Crowdsourcing a network status page On 7/12/2010, at 10:46 AM, Mauricio Freitas wrote:
About "expiring" we are only ever showing the last 72 hours on that page. Which I think is a good window size.
As for tests, I think we are not trying to create a sensor network, but crowdsource the information - this means what humans perceive as crappy services received, not what machines see as a fault...
72 hours is a very long time for expiration. You would only want 3-4 hours (maybe less), as otherwise people might get the impression the issue hasn't been fixed. Allowing a short time frame lets you offer more realtime status reports, two days is in computer land is a lifetime as some outages may only be for 15 minutes. Cheers, Drew Broadley drew(a)broadley.org.nz