PartitionKey, RowKey ordering

Question

I'm trying to decide on a new data schema at present. And I'm trying to remember....

If I make a query across multiple partitionKeys, then in what order is query data returned - is everything returned sorted (partitionkey, rowkey) or is a merge done so that everything is returned ordered by (rowkey) across all relevant partitions?

e.g. if I were writing a blogging platform and I gave each blogger their own partition, and used the smarx-inverse-ticks for rowkey, then would I be able to get back a time ordered list of most recent blogs across all partitions?

Thanks

Stuart (who really should just write the test for this!)

Answer 1

The following is documented:

Query results are sorted by PartitionKey, then by RowKey. Ordering results in any other way is not currently supported.

This suggests you will get time-ordered lists for each user.

Answer 2

Thanks

Yeah - I kind of remembered some of the details after I'd sent it - I think the query gets sent to one partition master at a time - and each one then either returns results or a continuation.

I'd forgotten how tricky it was to stop thinking relationally! I'll go look at things from a different angle.

Before I architect too much, I wonder if there is any news anywhere on when secondary indicies might arrive? I wonder if I have to wait for PDC...

Thanks again

Stuart

Answer 3

Before I architect too much, I wonder if there is any news anywhere on when secondary indicies might arrive? I wonder if I have to wait for PDC...

No. In my experience Microsoft has been reluctant to preannounce any Azure functionality and especially any date by which it would become available.