Thanks for your clarification Dean. But there are no plans at Google to support a relational persistent store?
I think it would be good if both are supported. For some applications I agree a relational database is not required, but for the kind of application I would like to write I think can't do without. With all those calculations it would get very messy. Also from my perspective I would not care about the fact that relationaldatabase queries will take more time and perhaps cpu usages. I'm willing to pay for it as my free quota is gone. I like to have more logic in the framework/relational database and less in my application. On 8 apr, 22:55, Dan Sanderson <dansander...@google.com> wrote: > I agree that the datastore is not good for adhoc runtime queries. The > datastore was designed to maintain its performance characteristics > independent of the number of entities stored, including query performance. > (The performance of our index-backed queries is a factor of the size of the > result set, not the size of the data set.) Consistent performance is > usually more important than adhoc runtime queries for large scale web > applications, so that's the focus of this design. > > I push the alternative techniques (e.g. calculating aggregates at write > time) because many web developers that grew up with single-server SQL > databases--myself included--are accustomed to pushing application logic into > database queries, because at a small scale that's often a best practice. > Some common tasks for web apps seem like they require SQL-style features, > but the lack of those features in a scalable environment doesn't make those > tasks unreasonable. Yet other tasks are genuinely untenable in a scalable > environment, at least not without heavy background processing. (I love our > new cron feature, but we're still looking at more thorough background > processing solutions.) > > -- Dan > > On Wed, Apr 8, 2009 at 1:34 PM, Marcel Overdijk > <marceloverd...@gmail.com>wrote: > > > > > > > The datastore implementation (non-relational) is one of the last > > things I'm bumping against (now Java is supported ;-) > > > I understand that aggregate data *can* be calculated at write time. > > But this is not what I want and I guess a lot of other users will > > think the same. > > > Also adhoc querying will be difficult with the current datastore. > > Maybe I want the sum of revenue per User > > or the sum of all users > > or maybe the sum per Country or, City, or... this would mean a lot of > > calculations to be done. IMHO this is also very error prone. > > > And maybe someday our business department wants average figures > > instead of sums... > > > I guess Amazon is not for nothing offering persistent storage which > > allows user to use e.g. MySQL. > > Also Sun's Cloud features a relational database (again MySQL). > > > I think 95% of all developers attracted to Google App Engine are using > > relational databases in the daily job... > > Like Andrew said it's a issue of barrier. > > > PS: I will look into the index documentation to see if it can help me > > with aggregate data. > > > Cheers, > > Marcel > > > On 8 apr, 21:40, Dan Sanderson <dansander...@google.com> wrote: > > > Some of the same problems can be solved in different ways. For instance, > > > aggregate data can often be calculated at write time, obviating the need > > for > > > an expensive aggregate runtime query involving millions of records and > > > hundreds of machines. The tricky bit is implementing the different > > > solutions using compatible APIs, which isn't always possible. > > > > -- Dan > > > > On Wed, Apr 8, 2009 at 12:31 PM, Marcel Overdijk > > > <marceloverd...@gmail.com>wrote: > > > > > Maybe for performance the datastore as it is now is best. > > > > But when working with data (e.g. aggregate functions like sum, avg > > > > etc.) a relational database has also advantages. > > > > > On 8 apr, 19:58, Andrew Badera <and...@badera.us> wrote: > > > > > It might not make "sence" but it certainly makes "sense" when you're > > > > living > > > > > in a world full of RDBMS, and want to make the barrier to entry as > > low as > > > > > possible. > > > > > > Thanks- > > > > > - Andy Badera > > > > > - and...@badera.us > > > > > - Google me:http://www.google.com/search?q=andrew+badera > > > > > > Sent from Albany, NY, United States > > > > > > On Wed, Apr 8, 2009 at 1:56 PM, Barry Hunter < > > > > barrybhun...@googlemail.com>wrote: > > > > > > > similar, but it wouldnt make sence to have two database backends.- > > > > Tekst uit oorspronkelijk bericht niet weergeven - > > > > > > - Tekst uit oorspronkelijk bericht weergeven -- Tekst uit > > oorspronkelijk bericht niet weergeven - > > > > - Tekst uit oorspronkelijk bericht weergeven -- Tekst uit oorspronkelijk > > > bericht niet weergeven - > > - Tekst uit oorspronkelijk bericht weergeven - --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Google App Engine" group. 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 -~----------~----~----~----~------~----~------~--~---
