As I was working with SQL Server 2014's new In-Memory OLTP database engine recently, I started to think about the big IT trends of the past 12 months. The In-Memory OLTP database engine itself will revolutionize how we think about memory and how we plan the implementation of our database servers. It provides a way to significantly increase database performance with little or no application changes by moving select tables completely into memory, with a new optimistic locking engine that processes queries directed toward those tables. However, because the In-Memory OLTP database engine is so new, it should still be considered a future technology.
Related: The In-Memory Revolution
Solid State Disks
Another technology that has clearly taken hold over the past 12 months is solid state disks. SSDs have reached critical mass in terms of capacity and affordability. In the past, SSDs just didn't have the capacity necessary for most production implementations, and the SSDs that did were too expensive to justify. This has all changed in the past year, as SSDs have become larger and the price has dropped. SSDs have now taken hold in a lot of SQL Server installations.
Multiple cores provide increased processing power. Many of today's tier-one servers support very large amounts of RAM, which increases their ability to handle heavy workloads. This has made the I/O subsystem the bottleneck for many workloads. Traditional hard disk drives have gotten larger, but they really haven't gotten any faster.
SSDs can be a real solution to this problem. For example, a 15,000rpm SAS drive can provide about 200MB/sec of throughput. In contrast, an SSD Serial ATA (SATA) drive using a 6GB connection can provide about 550MB/sec of sequential throughput.
Although there are no silver bullets for database performance (and SSDs are no exception), it's clear that when used properly with data, log files, indexes, or possibility tempdb, SSDs can provide significant performance improvements.
Of course, there are caveats to using SSDs. There are different types of SSDs, and SSDs have a shorter lifecycle than hard disk drives have. In an upcoming SQL Server Pro article, I'll cover SQL Server storage best practices, with greater detail about using SSDs with SQL Server.
Another important technology that has emerged in the past year is software-defined networking. SDN is essentially virtualization for the network.
Technologies have advanced to the point where server virtualization is now the standard. In addition, storage virtualization has become a reality. SANs have provided storage virtualization for some time, and the technology has now also surfaced in Windows Server 2012 in the form of storage pools. In each case, virtualization—which is essentially a software-based logical representation—abstracts the physical resource into a logical resource that the OS presents to users and applications.
SDN now brings this same capability to networks, with the end result of providing increased IT flexibility and faster application deployments. When a business deploys an application, it typically must set up a VLAN, which usually requires manual network configuration. If down the road the organization needs to move the application, this operation typically also requires manual network reconfiguration.
SDN and network virtualization allow you to create a logical network that utilizes but is independent from the underlying physical network. SDN lets you deploy applications with no physical network reconfiguration because policies associate the logical network with the application. Windows Server 2012 supports network virtualization, as does VMware's NSX. Check out my interview with Microsoft's Brad Anderson for more information on Windows Server 2012's SDN and network virtualization capabilities.
Up and Coming Trends
This leaves us with the trends that haven't yet taken off. Although most vendors, including Microsoft, are pushing hard to get businesses to adopt their cloud products, there are few takers—at least in the database industry. Concerns about security, trust, availability, and performance have kept most SQL Server instances firmly on-premises. The relational database is the heart of almost all business-critical applications, and businesses just don't want to take such a large risk.
A couple of areas where the cloud looks like it could make some inroads into the database world is for offsite backup and disaster recovery. IaaS and SQL Server virtual machines (VMs) could give some businesses that can't afford physical remote disaster recovery sites an affordable option.
Have I missed any big trends that you've seen? Drop me a line at www.sqlmag.com.