The Changing Future of Enterprise Storage


As you probably already know, enterprise computing is comprised of both applications and infrastructure. The applications are utilized for problem-solving, while the infrastructure is for hosting applications. Enterprise applications fall into eight different categories: business processing, IT infrastructure, and decision support systems like business intelligence and analytic, application development, collaborative, web infrastructure, industry R&D, high performance cluster computing (HPCC), and Internet computing.

Enterprise infrastructure consists of generations of heterogeneous architectures like physical and virtualization frameworks, centralized and distributed, dedicated and shared, segregated and converged and hybrid. An enterprise class solution is comprised of a set of application specific design concepts and selects HW/SW components. Each application’s infrastructure consists of server infrastructure like monolithic or blade infrastructure, management framework including in-band and out of band, interconnection technology, for example, Ethernet and Fibre Channel, application specific topology like fat tree, application HW/FW support matrix like switch and storage firmware versions, and application capability implementations for software level availability.

The IT professional’s job is to design enterprise computing operating environments – along with deploying, monitoring, managing and maintaining the applications and operating environment. Most enterprise class applications have specific requirements such as fabric design and response time, etc. For example, a mail server application may require 4 Ethernet fabrics: public LAN, private LAN, a storage area network (SAN) fabric and an optional fabric for backup. A high performance-computing cluster may require 56 Gbps Infiniband and non-blocking fabric. Mail servers may require guaranteed storage session response time such as 10 ms for writing and 20 ms for reading.

High-performance applications usually require a nanosecond-level interconnect latency rather than guaranteed storage read/write session time. Different applications can benefit from different storage architecture; for example, direct attached storage (DAS) provides best performance/cost ratio for certain mail servers, local storage and network attached storage (NAS) can satisfy most of the file/print server’s needs, SAN storage can meet requirements such as storage latency from Oracle applications, and virtual SAN storage such as Microsoft storage spaces can satisfy the requirements of certain virtual desktop infrastructure (VDI) applications.

Comments are closed.