On Wednesday, June 13, 2012 10:52:55 PM UTC+2, Jeff Schnitzer wrote: > > On Wed, Jun 13, 2012 at 11:33 AM, Brandon Wirtz <[email protected]> > wrote: > > I know exactly how much every request cost me. > > > > cpm_usd=0.000308 > > > > cpm_usd=0.000175 > > These numbers tell you how much a request cost, but don't tell you how > much another identical request will cost. > > Also, what do these numbers mean in the context of a multithreaded > instance? If you have 5 requests running in one instance and the 6th > request hits a fresh instance, how are the costs divided among the 6 > requests? > > > As to the latency changes in pricing are no different than when AWS has > a > > performance issue. (actually less so) > > In AWS, a performance degradation in the cluster may reduce your > application effectiveness, but it doesn't change billing - at least > not automatically. Maybe it will cause you to spin up new instances, > but this doesn't seem particularly likely - at least not in a > multithreaded world.
I'll have to disagree here. First off, you're talking about different "kind" of instances. On the other hand, if you're talking about some hypothetical app instances, you forgot to mention that someone (or something) will have to manage that. AWS is not running apps, it's running stuff like EC2 instances, which is a totally different thing. Secondly, for a mid to high load app you will definitely end up running more than one EC2 instance (otherwise, you might as well deploy your app on a PC sitting under your desk). If you don't setup Autoscale thing (if you do - here's your billing going up already), sooner or later you'll have to do it manually. Both ways, there will be additional costs. Not only that, you'll also need to allocate some (human) resources to manage that little cluster on which your app/whatever is running. GAE and AWS are not comparable, though they do have a couple similar services. > A java appserver can easily handle hundreds > (even thousands) of requests blocked on IO, so even under increased > "datastore" (lets say DynamoDB or SimpleDB) latency there's no reason > to expect a need for lots of extra instances. Theoretically, GAE's > multithreaded instances should work the same way... except that they > don't for some reason. > Jeff > On Wednesday, June 13, 2012 10:52:55 PM UTC+2, Jeff Schnitzer wrote: > > On Wed, Jun 13, 2012 at 11:33 AM, Brandon Wirtz <[email protected]> > wrote: > > I know exactly how much every request cost me. > > > > cpm_usd=0.000308 > > > > cpm_usd=0.000175 > > These numbers tell you how much a request cost, but don't tell you how > much another identical request will cost. > > Also, what do these numbers mean in the context of a multithreaded > instance? If you have 5 requests running in one instance and the 6th > request hits a fresh instance, how are the costs divided among the 6 > requests? > > > As to the latency changes in pricing are no different than when AWS has > a > > performance issue. (actually less so) > > In AWS, a performance degradation in the cluster may reduce your > application effectiveness, but it doesn't change billing - at least > not automatically. Maybe it will cause you to spin up new instances, > but this doesn't seem particularly likely - at least not in a > multithreaded world. A java appserver can easily handle hundreds > (even thousands) of requests blocked on IO, so even under increased > "datastore" (lets say DynamoDB or SimpleDB) latency there's no reason > to expect a need for lots of extra instances. Theoretically, GAE's > multithreaded instances should work the same way... except that they > don't for some reason. > > Jeff > -- 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/-/vPWCDJQVAUYJ. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
