Hi,
My favorite one at the moment is about using SQLite database stored in Azure BLOB, that stores indexes, and some schema-less values serialized by ProtoBuf.
This worked around slow SQL Azure and Table storage (lag problems), allowing to sync in multiple millions of products with arbitrary metadata per customer in matter of seconds. Random reads are really fast (esp. when SQLite file is cached locally).
It seems to be fun all the time))
Best regards,
Rinat Abdullin
Technology - Lokad.com | Blog - Abdullin.com |
+7 917 4613826