Big Dollars from Big Data
How to reduce costs and increase performance in the data center
Dec. 31, 2012 01:00 PM
Cloud computing has given birth to a broad range of online services. To maintain a competitive edge, service providers are taking a closer look at their Big Data storage infrastructure in an earnest attempt to improve performance and reduce costs.
Large enterprises hosting their own cloud servers are seeking ways to scale and improve performance while maintaining or lowering expenditures. If the status quo of scaling users and storage infrastructure is upheld, it will become increasingly difficult to maintain low cost cloud services, such as online account management or data storage. Service providers will face higher energy consumption in their data centers overall, and many are loath to begin charging for online account access.
Costs vs. Benefits
In response to the trend of growing online account activity, many service providers are transitioning their data centers to a centralized environment whereby data is stored in a single location and made accessible from any location via the Internet. Centralizing the equipment enables service providers to keep costs down while delivering improved Internet connections to their online users and realizing gains in performance and reliability.
Yet with these additional performance improvements, scalability becomes more arduous and cost-prohibitive. Improving functionality within a centralized data center requires the purchase of additional high-performance, specialized equipment, boosting costs and energy consumption that are challenging to control at scale. In an economy where large organizations are seeking cost-cutting measures from every angle, these added expenses are unacceptable.
More Servers, More Problems?
Once a telco moves into providing cloud-based services for its users, such as online account access and management, the demands on its data centers spike dramatically. While the typical employee user of a telco's or service provider's internal network requires high performance, these systems normally have fewer users and can access files directly through the network. Additionally, employees are typically accessing, sending and saving relatively low-volume files like documents and spreadsheets, using less storage capacity and alleviating performance load.
Outside the internal network environment, however, the service provider's cloud servers are being accessed simultaneously over the Internet by more users, which itself ends up becoming a performance bottleneck. Providers, telcos and other large enterprises offering cloud services therefore not only have to scale their storage systems to each additional user, but must also sustain performance across the combined users. Due to the significantly higher number of users utilizing online account tools at any given time, cloud users place a greater strain on data center resources.
Combining Best Practices
To remain competitive, cloud service providers must find a way to scale rapidly to accommodate the proliferating demand for more data storage. Service providers seeking data storage options should look for an optimal combination of performance, scalability and cost-effectiveness. The following best practices can help maximize data center ROI in an era of IT cutbacks:
- Pick commodity components: Low-energy hardware can make good business sense. Commodity hardware not only costs less, but also uses far less energy. This significantly reduces both setup and operating costs in one move.
- Look for distributed storage: Distributed storage presents the best way to build at scale even though the data center trend has been moving toward centralization. This is because there are now ways to increase performance at the software level that counterbalances the performance advantage of a centralized data storage approach.
- Avoid bottlenecks at all costs: A single point of entry becomes a performance bottleneck very easily. Adding caches to alleviate the bottleneck, as most data center architectures presently do, add cost and complexity to a system very quickly. On the other hand, a horizontally scalable system that distributes data among all nodes delivers a high level of redundancy.
Big Data storage consists mainly of high performance, vertically scaled storage systems. Since these current architectures can only scale to a single petabyte and are expensive, they are not cost-effective or sustainable in the long run. Moving to a horizontally scaled data storage model that distributes data evenly onto low-energy hardware can reduce costs and increase performance in the Cloud. With these insights, providers of cloud services can take steps to improve the performance, scalability and efficiency of their data storage centers.