Sequence object
Produce numbers that can be used to populate many tables instead of identity that only applies to 1 table.
http://www.databasejournal.com/features/mssql/sql-server-denali-sequence-object.html
CREATE SEQUENCE MySequence
AS INT
START WITH 1
INCREMENT BY 1
MINVALUE 1
MAXVALUE 10000
CYCLE
CACHE 50;
GO
SELECT NEXT VALUE FOR MySequence
Columnstore Indexes¶
Column data is stored in the same pages per column. This can then be compressed due to data replication.
Good for fact tables. Will improve ROLAP. Readonly.
Secondary index - original rowstore indexes must also exist so storage is a downside.
If dimension tables have > 10 mill rows then maybe use column store on there as well.
Imprevements will be most where column store will fit in memory but rowstore will not.
Up to 1000x perf increase is possible.
Can maintain 2 tables - 1 with columnstore (readonly) and 1 with new rows fed in from oltp (deltas table). Queries can be modified to access both tables. Lets create a VIEW over both tables you say...well no -
batch-mode query execution doesn't work consistently across UNION ALL (yet), so creating a view combining the columnstore table and delta table and querying the view won't work
A possibly nightly trickle load can be introduced to feed the columnstore from this secondary "differential" table.
Read only but index can be disabled - data updated and then index rebuilt. Problem is that process can't be done just on a partition, has to be the entire index - BOO!Always ON
New mirroring. Instead of being limited to 1 mirror it can be > 1.
FileTable
Bring files (that contain non organised\non column style data) on the OS that are referenced by apps inside sql server. Maintains heirarchical storage of folder structure. Gain admin\security benefits of sql server.
Supports full text and semantic search in the files.
Built on-top of FILESTREAM from 2008.
Partitions
1,000 to 15,000 partitions means 3 years of daily partitions become 45 years.
Power View
Replaces report builder. Can be used on win 8 and ipad.
Runs in the browser and is working against a BI Semantic Model (BISM). BI Semantic Model is available on the client through the Excel based Power pivot modeling tool or on the server through Analysis Services Tabular project.
Juneau
Database project in vis stud\bids. Actually really cool in terms of auto debug database on sql express, refactoring supported by a change log that is auto popped in the background.
Auto change script creation.
Promotion of change scripts into the debug database.
Schema compare tools.
Intellisense upgrades.
Deployment reports generated to tell you what will be rebuilt, what code will break due to object drops etc...
LocalDB (express) is installed with Juneau as a debugging platform for local t-sql. A new instance of express is created for every project in a database solution file. They have their own logs etc and will close down if not used to free up resources.
Distributed Replay
Distributed Replay offers a more scalable solution than SQL Server Profiler. When multiple concurrent connections are interacting with the db profiler may become a bottleneck when replaying.
Replay has a command line tool and a service to play back from.