The ongoing proliferation of information and the need to digitally catalogue all of it has resulted in a massive glut of data. The challenge for users is to figure out how they can best store and managethis huge amount of data.
Statistics from Gartner show that in 2008, for the first time, the amount of file-based storage system capacity shipped surpassed block-based storage system capacity shipped. Based on this trend, it is estimated that file-based storage system capacity will account for 70 percent of total storage capacity by 2012. Additionally, IDC predicted that in view of the rapid growth of file-based unstructured data, 80 percent of the global storage market’s total shipments will be covered by file-level data by 2012.
Traditional NAS systems
Traditional network attached storage (NAS) systems are now encountering a bottleneck in the face of the creation of these massive amounts of data. This bottleneck can be attributed to three main disadvantages that hinder traditional NAS systems.
First is low controller utilisation and space utilisation. In traditional NAS equipment, each NAS controller has its own separate storagespace which results in the low utilisation rate of the controller since data cannot be shared between controllers.
The performance of a single NAS controller is limited in scope and there exists a hot application access bottleneck. For instance, in scenarios such as high-performance computing, a computing task can only be undertaken by one NAS controller. Other controllers cannot share the workload, leading to wasted resources and computing inefficiencies.
The second disadvantage is low scalability. Traditional NAS systems are single/dual controller architectures, which, by design, have limited scalability. Performance of traditional NAS systems can only be enhanced byactually increasing sets of equipment. However, rather ironically, storage resources cannot be shared between different sets of equipment, so this method actually increases the management complexity of data since it creates disparate data islands.
The third disadvantage is management complexity and high maintenance cost. An increasing amount of users and enterprises own dozens or even hundreds of file servers or NAS devices. With each device independent from one other, this situation creates a scattered array of devices which results in increased management complexity and higher maintenance costs.
The emergence of clustered NAS systems
With an increasing amount of user data, IT systems have become more complex. Given this situation, clustered NAS systems have gained popularity among users over the use of traditional NAS devices. Huawei Symantec, for instance, offers the N8500, a high-end clustered NAS product that was recently recognised as having achieved the industry’s highest results on the Standard Performance Evaluation Corporation’s (SPEC) SPECsfs2008 test.
Clustered NAS systems offer users an array of advantages over traditional NAS systems. The first advantage is that clustered NAS systems allow for higher controller utilisation and space utilisation. All NAS controllers of a clustered NAS system have back-end storage space where data can be fully shared. With all storage space integrated into a larger resource pool whereresources can be allocated based on actual needs, space utilisation is thereby enhanced. In addition, all NAS controllers can work on the same task, dramatically increasing the utilisation of each controller.
Clustered NAS systems have absolute advantages, especially for high-bandwidth applications such as video on demand. Access to videos that are in high demand can be achieved through multiple controllers, taking advantage of the system’s ability to engage in enhanced parallel processing andeliminating bottlenecks created through the ongoing access of high-demand data.
The second advantage is that clustered NAS systems can achieve higher degrees of scalability. For instance, clustered NAS systems can be extended from two online dimensions by extending the NAS controller to meet the growing demand for performance or by extending the storage capacity to meet the data storage space requirement. In a situation that requires large-capacity backup, expansion of the back-end storage unit would allow a user to meet the growing demand for capacity.
Additionally, in high-performance scenarios, the expansion of NAS controller node would enhance the processing capacity necessary to allow for concurrent user access. A two-dimensional expansion enables the whole system to be more flexible and more adaptable.
Finally, clustered NAS systems lay the groundwork for simplified system management as well as for reduced operating costs. In light of the high performance and scalability of clustered NAS systems, a single system can easily provide a GB/s of bandwidth and a PB or more of storage capacity. Given these high storage and transfer rates, a single easily managed and maintained set of equipment is sufficient to meet any user’s data storage needs, no matter how extensive.
Huawei Symantec’s N8500 combines the aforementioned advantages with additional benefits such as a faster rate of response, easy storage expandability to cater to surges in business growth, and a simplified IT structure due to the ability of the N8500 to meet the storage needs of an entire IT system.
In developing the N8500, Huawei Symantec worked with certain universal parameters in mind that would be ideal in any clustered NAS system. Several important features were included in the development process including file-based automated tiered storage, snapshots, data replication, and file data. Understanding that users will be drawn from a wide range of industries including telecommunications, digital media, high-performance computing, government, and education, the N8500 was developed with a comprehensive feature set, high performance, scalability and efficient data management.
By Shabir Satar, product manager, Huawei Symantec