Linear scalability is essential for attractive TCO and ease of management. Legacy NAS is not architected for efficient linear scaling, making it cost prohibitive to reach sufficiently high performance levels in a single high capacity system. Partitioning legacy NAS into multiple smaller systems can sometimes be a solution, but at the cost of additional management effort for each new NAS island and data partition. Clustered NAS head architectures can ease management effort, but in-band metadata processing and coordination between clustered NAS heads impact scaling efficiency. With legacy NAS, it is common to experience only a fraction of the performance of each clustered head at scale — impeding single system deployment with prohibitive cost.
True linear scalability delivers a low Total Cost of Ownership (TCO) at scale without the compromise of high management effort. In the end, linear scalability ultimately drives single system maximum performance. Maximum performance drives faster time-to-results and accelerates workflows. Innovation and discovery are unlocked as business organizations become more competitive and profitable.
ActiveStor®—Designed for Linear Scalability
ActiveStor® is architected for linear scalability to exceptionally high performance levels. The PanFS® storage operating system at the heart of the ActiveStor solution ensures efficient linear performance with scale:
Many legacy and competitive NAS systems claim linear scalability, but cannot demonstrate scaling efficiency at massive scale. ActiveStor demonstrates linear scalability in real-world deployments with over one hundred ActiveStor shelves and more than twelve petabytes of capacity. The true price for a given level of storage performance is only seen at desired scale where linear scalability is often one of the largest TCO factors. Panasas customers around the globe have chosen ActiveStor not only for exceptional performance, manageability, and reliability — but also for the linear scalability that enables them to achieve a compelling price/performance at scale.
Technical research and enterprise customers are processing large, rapidly growing quantities of data to drive innovation and discovery. Scale-out NAS promises a large single global namespace with efficient performance scalability to accelerate workflows and time-to-results. However, this promise is not without significant challenges, especially at large scale.
True linear scalability requires an efficient storage architecture. Efficiency means that a doubling of capacity doubles performance for large file and small file workloads — regardless of scale. While it is relatively easy for a single legacy NAS-head product to double performance with an additional clustered head, the real scale-out NAS challenge is to reach 20 times the performance from 20 NAS heads. Legacy NAS architectures hit a scalability "brick wall" in which only a fraction of the performance of an added head is delivered at scale.
While all scale-out NAS solutions scale to some degree, the challenge for technical research and enterprise environments is to seamlessly and linearly scale capacity and performance within a single storage system — without adversely impacting manageability and reliability in the process.