Condusiv Technologies, the leader in high-performance software optimizing technology, people and businesses, today is announcing the results of strict, third party benchmark testing of its newly-released V-locity Server Optimization software designed for I/O-intensive applications like SQL Server and Exchange running on physical servers.
Condusiv’s V-locity software architecture, which contains transformational read and write optimization engines, represents the culmination of 31 years of research and development in optimizing and accelerating Windows environments for business.
Condusiv’s goal is to help broaden industry awareness of the benefits of V-locity’s unique approach to optimizing read/write performance at the source, addressing critical I/O performance barriers without adding storage or server hardware. Condusiv Technologies is presenting a sponsored Technology Spotlight by leading IT market research and advisory firm IDC, entitled “The Shift to I/O Optimization to Boost Virtual and Physical Server Performance.” In addition, Westborough, Massachusetts-based openBench Labs released a third party test report revealing that V-locity Server accelerated SQL performance by 55%. Access both papers at http://www.condusiv.com/business/v-locity/server/.
“Virtual environments, cloud services, mobile devices, and Big Data all contribute to the rise in digital information organizations must manage. All of this data not only must be stored, but utilized to drive value for an organization's competitive advantage,” said Jerry Baldwin, CEO of Condusiv Technologies. “As much as the I/O explosion needs to be managed, CIO’s find themselves investing 80% of their annual IT budget on maintaining their existing infrastructure and services. This model is broken. V-locity customers typically see 50% or more performance gains on mission-critical applications like SQL and Exchange. That gain also comes with a very unique proposition—a savings of about 80% from their annual hardware capital expense budget.”
The Problem
The I/O problem stems, in part, from the fact that while the number of virtual machine shipments is growing at an average of 25% annually, the number of physical servers shipped is growing at a modest 2–3%. As more workloads are put on virtual servers and heavier workloads are placed on physical servers, this can triple or quadruple the amount of random I/O generated from a single server, burdening the compute infrastructure. Increasingly, the storage controller and disk architectures cannot keep pace with this growing random I/O.
When it comes to I/O and its impact on servers, storage and applications, there are two performance barriers: 1) Windows creating unnecessary I/O traffic by splitting files upon write, which also impacts subsequent reads, and 2) frequently accessed data unnecessarily traveling the full distance from server to storage and back.
These two behaviors create a surplus of I/O that prevents applications from performing at peak speeds. In today's enterprise, the problem is compounded as a multitude of random I/O traffic, from a mass of disassociated data access points, is making requests for storage blocks—random and sequential—to a shared storage system. All this unnecessary I/O leads to extra processing cycles that increase overhead and reduce application, network, and storage performance.
The I/O problem will continue to grow. IDC predicts the amount of information that needs to be managed by enterprises will increase 50 times in the next 10 years, and the number of files will increase 75 times. However, with Moore's law slowing from processor speeds doubling every 18 months to doubling every three years, processor performance will grow only by a factor of eight and storage performance will grow by a factor of four1.
Condusiv’s V-locity Server optimization software addresses critical I/O issues by eliminating application bottlenecks without the need to add server or storage hardware. Condusiv's differentiator is that its software resides at the top of the technology stack, eliminating unnecessary I/O at the source, where it originates.
As a first step to I/O optimization, V-locity Server eliminates nearly all unnecessary I/O operations at the operating system level when writing a file, which in turn eliminates all unnecessary I/O operations on subsequent reads. Second, V-locity Server caches frequently accessed data within available server memory without resource contention to the application to keep read requests from traveling the full distance to storage and back.
With V-locity at the top of the technology stack, optimizing I/O at the point of origin, this means only productive I/O is pushed through the server, network and storage. This approach to I/O optimization complements technologies that may already be running to promote IOPS or latency reduction, including SSDs, flash cards, and SAS, and provides tremendous benefit from the top-down. And since I/O is optimized at the source, V-locity Sever is network storage-agnostic, providing benefits to advanced storage features like snapshots, replication, thin provisioning and deduplication.
Solution: V-locity Server increased SQL Server 2012 transaction processing rate by 55% and improved response time by 33% without additional hardware.
openBench Labs tested the ability of V-locity Server to optimize I/O in a SQL Server environment. Using SQL Server 2012, openBench tested a mix of a high volume of light-weight SQL select transaction processing (TP) queries, combined with heavy-weight background update queries.
For the SQL Server benchmark testing, openBench simulated 1 to 32 daemon processes (1 daemon generating the equivalent of 70 normally-queued user processes) issuing queries non-stop. When a real application user interacts with SQL Server, there is lag between queries issued. In the test scenario, however, the daemon process issued queries without lag—that is, no think-time, type-time, or pause-time between query activity.
In a series of tests, openBench Labs measured the ability of V-locity Server’s IntelliMemory™ to offload I/O on read operations through dynamic caching in order to boost throughput and reduce latency. In addition, openBench examined the ability of IntelliWrite® technology to prevent unnecessary split I/Os, using its intelligence to extend current database files and create new log files as single, contiguous collections of logical blocks.
In a test of SQL Server query processing, openBench Labs benchmark findings revealed that V-locity, on a server running SQL Server, enabled higher transaction per second (TPS) rates and improved response time by reducing I/O processing on storage devices. What’s more, in a SAN- or NAS-based storage environment, V-locity Server reduced I/O stress on multiple systems sharing storage resources. Overall, V-locity Server can improve scalability by reducing average response time and enabling SQL Server to support more users.
“On a server running SQL Server 2012, V-locity Server created an environment that enabled up to 55% higher TPS rates, improved transaction response time by 33%, and enabled SQL to process 62% more transactions at peak transaction rates. As a result, IT has a powerful tool to maximize the ROI associated with any business application initiative driven by SQL Server at the back end,” said Dr. Jack Fegreus, founder of openBench labs.