All companies rely on some critical applications in order to conduct their business. When these critical applications perform poorly, there are a variety of negative consequences to the business including poor customer service, lost revenue, decreased employee productivity and frustrated users. To illustrate this point, envision company service reps sitting idle waiting for a customer relationship management (CRM) application to retrieve the vital records required to service customers.
What is the root cause of these critical applications performing poorly? The root cause of nearly 80 percent of poor application performance can be traced back to storage input/output (I/O) delays as a result of disk I/O bottlenecks. These I/O bottlenecks are a result of disk latency and directly linked to the company’s data storage infrastructure.
While hard disk drives (HDD) historically have been a common storage technology, this type of storage is mechanical in nature and is inherently slow.
A faster alternative is solid state device (SSD) storage that delivers improved performance over HDD. However, within the SSD category, there are different types of storage technologies. The first is Flash SSD, which is highly affordable yet has inconsistent performance for read/write operations and some architectural reliability issues. A second type, DRAM-based SSD, is both fast and highly reliable, though it requires a redundant UPS to fully protect data in the event of power loss.
Of the two SSD technologies mentioned, DRAM-based disk technology can offer significant improvements in speeding critical business applications and ensuring data resilency and reliability. Here are some real-world examples of how well DRAM storage solutions can increase application performance:
To achieve this level of application performance improvement, IT organizations should look for the following characteristics:
High Performance: The solution should be able to eliminate I/O bottlenecks and capable of delivering up to over one million IOPs and up to 16 GBs of throughput with the same sub millisecond latencies. Consistent performance is also important as workloads vary and dynamic load balancing is also critical to preventing hot spots. The ideal solution should not see performance degradation over time or require any performance tuning.
High Availability: The solution should not have a single point of failure. It should support automated hardware and data recovery. Critical business data must be fully available and reliably maintained with integrity through internal data protection and dual universal power supplies.
Scalability: As the volume of critical data grows over time, companies need a scalable storage solution. Preferably, the solution should be based on standard, high-end blade components which easily lend themselves to modular scalability. Throughput and capacity should be independently scalable as needs often change over time.
Easy to Install: The solution should be easy to install, requiring minimal time and effort from the IT organization. Optimally, the storage solution should be transparent and not require any modification or disruption to applications or corporate data.
This optimal solution is currently available from Kaminario with its Kaminario K2 DRAM-based storage appliance. This field-tested solution has been used in telecom, financial services, e-commerce, and healthcare environments and has delivered extreme I/O performance improvements with millions of I/O per second (IOPS). The Kaminario K2 has dramatically reduced application response time in a number of critical applications within businesses of all sizes.
An investment in DRAM-based storage solutions to solve the I/O bottleneck problem can pay huge dividends in a company’s operations benefitting customers, employees and IT departments.
Eyal Markovich is a seasoned performance subject matter expert and leads the performance team for Kaminario.