Transforming Internet Datacenters
Using fully optimized data access appliances
Jun. 9, 2009 12:45 PM
With the explosive growth of Web-based businesses and applications, datacenter workloads have increased exponentially. IT managers are finding it difficult to meet the accelerating demands for performance, capacity, scalability and reliability, while at the same time meeting budgets, maintaining service-level agreements and driving green initiatives.
New component and middleware technologies - including multi-core processors, low-latency interconnect, flash memory, and highly optimized data access and caching technologies - hold serious promise for accomplishing these goals. But without optimal integration and implementation, these technologies have failed to deliver the benefits that datacenter managers and the business demand. Because Web 2.0 and cloud computing enterprises must focus on their core business, higher-level building blocks are needed that can exploit these advanced technologies to fundamentally solve these costly challenges.
Datacenter Trends and Challenges
With the explosive growth of Web 2.0, software-as-a-service (SaaS), cloud computing and other emerging Web-based applications, datacenter workloads have increased exponentially. The business opportunities that are created from the deployment of these new applications are substantial, but the demands they place on the datacenter are daunting. Challenges include:
- Unprecedented data growth. Recent studies indicate that the amount of data managed by today's datacenters will quadruple every 18 months. To complicate matters, online users are getting much more sophisticated, and response time expectations are at an all time high. But with the large increases in data volume, user interaction times are actually increasing for many datacenters.
- Severe capacity constraints. Datacenter managers are struggling to manage huge increases in rack, power and network utilization. They are constrained by limited datacenter power and space and are seeking cost-effective ways to expand capacity without increasing the datacenter footprint.
- Increasing data complexity. Organizations have too much data to process in a time-sensitive and consistent manner. Information management requires extensive data partitioning and application-level mapping, caching, replication/recovery and load balancing. Existing data management tools are complex and existing commodity, non-application-specific hardware is difficult to use and manage.
- Lack of scalability. Current datacenter environments lack the ability to scale effectively to manage peak demand. Even in multi-core server environments, many datacenters have already scaled to the point they are memory and disk bound.
- Underutilized resources. Estimates of current datacenter equipment utilization rates run between five and 30 percent. Kilowatt-hours are very expensive and inefficient. Underutilized hardware translates into wasted capex and opex, as well as increased power consumption.
- Severe budget constraints. Budgets are tighter than ever. Only solutions that are able to provide quick return on investment (ROI) are now being approved. Traditional approaches to datacenter expansion are no longer viable.
- Corporate mandates to go green. According to the 2007 EPA Report to Congress on Server and Datacenter Energy Efficiency, the energy consumption of servers and datacenters has doubled in the past five years and is expected to almost double again in the next five years to more than 100 billion kWh, costing about $7.4 billion annually. Datacenter managers are working hard to save energy, reduce datacenter space requirements and protect the environment.
Limitations of Today's Technology
The ability to effectively scale the database and caching tiers is critical for the success of any growing Web-enabled business. Unfortunately, ordinary server and middleware infrastructure is loosely integrated and minimally optimized. Existing datacenter solutions aren't adequately addressing the performance, capacity, scaling, reliability and power challenges of supporting new Web applications effectively.
Several new technologies have the potential to solve these challenges:
- Multi-core processors offer the promise of improved application performance and increased energy efficiency, but harnessing that potential effectively requires very high parallelism and synchronizing the processors, with a resultant higher degree of complexity.
- Enterprise-class flash memory has several advantages over traditional hard disk drives (HDDs), including 100 times faster access time, increased reliability due to no moving parts, and significantly reduced power. Space and power savings, for the same server workload, translate into a tangible reduction in total cost of ownership (TCO).
- Low-latency interconnect technologies provide the ability to send a message in a millionth of a second - more than an order of magnitude improvement over existing solutions.
- Optimized data access and caching applications can speed access to data. But these applications need to be deeply integrated with the operating environment and hardware to fully utilize the benefits of multi-core processor, flash memory and low-latency interconnect technologies.
The unfortunate reality is that each customer's IT team must become its own integration house, developing software in an attempt to optimize these advanced technologies - a highly complex and error-prone process that can take years to implement. This approach not only diverts valuable IT resources from the enterprise's core business, it results in unmanageable application and administrative complexity, low-resource utilization and high TCO.
Challenges of Utilizing New Technologies
The new generation of commodity multi-core processors, flash memory and low-latency interconnects offer tremendous potential in Web 2.0 and cloud computing datacenters, but in reality, due to the extensive work to implement, the benefits are limited. The effort required to utilize these new technologies to solve today's severe performance, power, space and TCO challenges is significant. IT teams need to develop highly parallel middleware applications, a high-performance operating system, and develop and optimize numerous, specialized configurations.
Adapting or inventing new deployment architectures to take advantage of the new technologies is a major undertaking, with large development and support costs that are not the core value of the businesses. Fortunately new, higher-level building blocks are now being introduced that address these challenges.
New technology integrates advances in flash memory, multi-core processors, low-latency interconnect, and optimized data access and caching technologies into fully integrated, optimized, multi-node appliances for faster data access. In the first half of 2009, several vendors introduced all-new, fully integrated appliances that leverage some or all of these technology advances.
By using these fully optimized data access appliances, enterprises can leverage higher-level building blocks that eliminate the need for complex integration projects. These appliances enable IT to meet corporate goals of increasing capacity, scalability and reliability, while at the same time slashing costs and dramatically curtailing energy consumption across the datacenter. The net result is not just more efficient datacenter operations, but the creation of new revenue-producing business opportunities based on rapid access to terabyte-scale data.