Yes, we'd love to do that at some point in the future - pick a general area where your apps would exist, and we'd run them from those nodes. I suspect that if we ever did this feature and you wanted to move to a new cluster, that you would have to run the datastore migration tool (similar to what it would take to move from MS -> HR), and if you're going to do this, why not just move to high replication?
When wooble mentions a single Bigtable, what he means is a cluster of Bigtable nodes, which consists of data stored on multiple Bigtable tablets. A "single Bigtable" is just the way we logically talk about Bigtables: http://code.google.com/appengine/articles/storage_breakdown.html The silver lining here about these two maintenance periods is that the next maintenance period should be far off. The reason we needed to do these read-only periods won't resurface anytime soon. Since I know somebody is going to ask: I'm not 100% how long before we have to do another maintenance period. My guess is a few months. Keep an eye on the downtime notify list. Ikai Lan Developer Programs Engineer, Google App Engine Blog: http://googleappengine.blogspot.com Twitter: http://twitter.com/app_engine Reddit: http://www.reddit.com/r/appengine On Thu, Apr 7, 2011 at 3:52 PM, László Török <[email protected]> wrote: > > > On Apr 7, 4:41 am, László Török <[email protected]> wrote: >> > This is a good idea and my gut feeling is telling me it should be >> possible. >> > You could basically specify your apps geographical availability (say by >> > continent) and Google could group apps onto a set of nodes using these >> > hints. >> > Maintenance would take place by "regions". I'm not sure how much control >> the >> > GAE folks have over the underlying storage infrastructure though. >> >> Considering all of the data for the whole master-slave datastore is >> believed to be stored in a single BigTable in a single datacenter >> somewhere in California, this could be tricky. >> > > That doesn't say anything about how that "single" BigTable is distributed > across nodes. > I've have seen reports stating that update operations failed on a certain > number of entity groups. > This suggests some nodes were failing on which those entity groups where > stored. > > This nothing more, just speculation as I'm not on Google's engineering > team, but thinking of a single BigTable as one monolithic entity is naive... > :) > > Maybe someone from the GAE Team can shed light on this matter. > >> >> -- >> >> You received this message because you are subscribed to the Google Groups >> "Google App Engine" group. >> 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. >> >> > > -- > You received this message because you are subscribed to the Google Groups > "Google App Engine" group. > 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. > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. 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.
