I don't understand why people keep adding DBA or systems engineer costs as a benefit of appengine. Most startups don't need those profiles at the beginning. Most engineers can do that work at design time. It's not that difficult to setup a fail-safe cluster.
I have done all of those optimizations you are talking and many more, like the most crazy things: minifying the JAR's with proguard so classpath scanning is faster and boot time is reduced. Again, I love appengine and have been using it since day one, but it's expensive. It looks like you have a lot of experience on appengine. Please, be kind share your experiences you hit 1 million users access your site every day. It will be very insightful. thanks On Wed, Jan 22, 2014 at 6:14 PM, Andrei Volgin <vol...@spiraluniverse.com>wrote: > > App Engine is very cheap for startups. I would even say it's a > game-changer. I built two startups before PaaS became an option, and I can > say with certainty that App Engine would have saved me millions of dollars > each time. > > Most new applications can happily live within a free quota until they > prove to be useful for a significant number of users. A well-built app will > have a to serve a whole lot of users before it costs more than $200 per day > on App Engine, and that's still cheaper than a single systems engineer or a > DBA. > > I think that App Engine is difficult and expensive for less-experienced or > less-qualified developers. Sloppy data models, unnecessary indexes, overly > complex queries, unintended dependencies, large third-party libraries for > simple tasks, wrong headers on static resources - these mistakes are not > always noticeable when you run your own box, given how cheap and powerful > the servers are these days. With App Engine these mistakes add up very > quickly. > > On the other hand, App Engine gives you all the information you need to > analyze your bill. Once you see that a certain item becomes significant, > there are ways to optimize your application in order to reduce costs. I > certainly advocate avoiding basic mistakes from day one, but there are > certain optimizations that make sense only when volume picks up. For > example, you can move a backend task to a Compute Engine instance, which is > many times cheaper, but requires more work to set up and manage. Or you can > split a complex data entity into two separate entities so that a minor > change will not result in multiple datastore writes. Or you can unindex > some properties and iterate through query results, saving on writes and > data volume. Or you can set the correct chunk size on your queries - > something that many developers probably forget to do. And so on. > > I am working on an app now that loads, processes and indexes 1 million web > pages and creates app. 10 million datastore entities for less than $100 in > App Engine costs. I don't know what your application does for your users, > but that's a whole lot of processing power for a hundred bucks. Once we hit > a million users or so, we will move some of our processing load to the > Compute Engine, but the effort does not make economic sense before that. > > -- > You received this message because you are subscribed to the Google Groups > "Google App Engine" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to google-appengine+unsubscr...@googlegroups.com. > To post to this group, send email to google-appengine@googlegroups.com. > Visit this group at http://groups.google.com/group/google-appengine. > For more options, visit https://groups.google.com/groups/opt_out. > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To unsubscribe from this group and stop receiving emails from it, send an email to google-appengine+unsubscr...@googlegroups.com. To post to this group, send email to google-appengine@googlegroups.com. Visit this group at http://groups.google.com/group/google-appengine. For more options, visit https://groups.google.com/groups/opt_out.
