OS Challenges for Data Management

Demand for big data delivered at faster speeds, driven by applications such as smartphones and analytics, is fueling an industry boom. Venture capitalist Ann Winblad famously called this boom "the new oil," sparking controversy, according to Forbes. While Winblad's analogy may be debated, big data's impact on technology and the economy is clear, with firms scrambling to find workers sufficiently skilled to manage the information explosion, as CNN Tech points out. This puts database administrators in the enviable position of occupying one of today's fastest-growing jobs. But it also forces them to keep up with how new data management methods interface with operating systems.

Operating System Issues for Database Management

Operating systems and database programs can suffer a failure to communicate, a problem analyzed in a seminal paper by computer scientist Michael Stonebreaker, as Gigaom notes. Stonebreaker is the developer of Ingres and PostgreSQL and a critic of NoSQL trends. Differences among programs can arise over numerous areas, including management of buffer pools, file systems, scheduling, processes, inter-process communication, consistency control, and paged virtual memory.
When such conflicts become acute, operating system data processing requests can drag and slow database programs. This presents a challenge for operating system designers, as well as database administrators. Network administrators in charge of services such as VPS Windows solutions are also affected by how network operating systems interface with databases they support.

SQL Challenges

Until recently, most database programs adopted an SQL approach, and supported corresponding operating systems. The most popular SQL database management systems include Oracle Database, Microsoft SQL Server, MySQL, PostgreSQL, and IBM DB2, according to DZone. These options are generally compatible with common operating systems, including UNIX, Linux, Windows, and Mac OS X, with Linux preferred among professionals, as Computer Weekly reveals.
But traditional SQL faces a limitation when handling big data. SQL sacrifices speed for accuracy, complying with what industry jargon calls the ACID test (Atomicity Consistency Isolation Durability). This ensures that vital data is not copied over or lost during a system crash. In order to speed up enough to handle big data, SQL must loosen ACID rigor, potentially risking data. A high-profile example, illustrated in a post on Infoworld.com, is Facebook's proneness to stall because its MySQL data management system cannot handle its huge workload.

The NoSQL Solution

To address this, some database designers have developed NoSQL programs that trade the relational element of SQL for scale and speed. Popular examples include MongoDB, Cassandra and Redis. These generally provide cross-platform support. MongoDB, for instance, works with Linux, Windows, Mac OS X, and others, according to it's site.
However, NoSQL databases cannot guarantee ACID compliance. Critics such as Stonebreaker complain that this sacrifice comes at the price of only marginal gains in speed.

The NewSQL Alternative

Stonebreaker and like-minded developers have proposed an approach to database management that combines ACID accuracy with NoSQL scale and speed, a synthesis they call "NewSQL." An example of NewSQL is VoltDB, an in-memory database developed by Stonebreaker and others. VoltDB requires a 64-bit Linux-based operating system and officially supports CentOS 5.8 and 6.3, RHEL 5.8 and 6.3, and Ubuntu 10.4 and 12.4. It also offers development builds tested to work with Mac OS X 10.6, as outlined on Voltdb.com.