The NoSQL movement is pushing the threshold ahead in cloud based data management and storage to support the exponentially increasing massive data requirements we are seeing in the online world today. Yes, these technologies do not currently address all the ACID properties that RDBMS products support.
Yes, these products are currently startup, incubator based, open source technologies and initiatives that are not currently standards based but I envision that these issues WILL be solved shortly and minimally these cloud based technologies can be in the short term integrated into portions of relational products that address certain performance critical areas such as data movement and caching. Oracle just announced BDA (Big Data Appliance) last week so this is already given.
I see GIS with immense amounts of data as having one of the biggest needs for this. Some of the data paradigms involved like graph databases are a perfect fit for GIS and especially ESRI products in general. I would like to see a large GIS data project with a Hadoop type of map-reduce solution for the data movement portion along with a traditional relational transaction/business logic architecture for a backend geodatabase.
