As I have been reviewing new features being built into SQL Server 2016, I found one I can really appreciate. Stretch Databases. Using stretch database, your data is partitioned according to frequency of access. Data used less frequently is archived onto different storage, isolated from the highly accessed data. The archived storage leverages Azure for reasonable performance at a reduced cost.
Many companies pay large sums for a similar appliance, usually built into a single SAN, having different storage devices with different performance characteristics and costs. Data used most frequently is on high performance drive configurations, or even SSD. Less used data may be stored on a RAID 5 set of spindles. Old data, rarely used, may be stored on inexpensive, lower speed drives, etc.
Stretch databases provide more power for the database with different data access patterns. It also leverages performance in a different way. Rather than slowing down drives, it distributes the data. So, slowness in processing is due to network latency, rather than the speed of the storage device. Much like partitioning, data stored on azure may be queried locally, and return only the results of the query, radically improving performance when addressing large volumes of data.
Anyone who has worked with data for a short time generally runs into a wall of database storage resources. We run out of disk space, and have to make some choices. Delete data. Archive data. Compress data. Purchase more disk. With Stretch databases, you can add it to your existing database, perhaps with less pain than the other options you had to use previously.
I’m sure I want to test this further. This is a feature I can see becoming very popular.
Cheers,
Ben