Hi I thought I would chime in, I am slowly starting to get my head around what you are suggesting in terms of scheduler behaviour.
Though you do mention min idle-instances (which isn't a tuneable parameter as I understand it.) So I will mention what I am seeing on a low frequency use site. We are running always on instances. Code hasn't changed in an appreciable way over the last few months. (changed analytics to async loading and a couple of cosmetic changes) Webmaster tools reports that for all of June my site latency was down in the sub 3 seconds. As of the 1.5.2 update it has been steadily climbed to 6 secs. Its more than likely because of out low traffic (about 200 page views a day) that someone will always hit an idle always on instance. All of the dashboards configurations are currently set to default (max idle auto, min pending latency auto) So it would seem on the face of it, that the introduction of the new regime (though incomplete) that default settings are less than optimal than pre-1.5.2 and because there are no real statistics presented in this space the only way to try and remedy the situation is trial and error, which isn't really a good option on a live site. Are you going to try an compile/present some sort of statistics so that we can better understand what is going on. Like a instance miss rate, % requests not server by always on/idle, rate of new instance creation etc... Thanks Tim -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To view this discussion on the web visit https://groups.google.com/d/msg/google-appengine/-/9OS1oamAV2wJ. To post to this group, send email to google-appengine@googlegroups.com. To unsubscribe from this group, send email to google-appengine+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
