Chapter 17: Defining the IO Workload

Overview

Capacity planning has always been something of a black art. Commodity-level distributed computing has driven IT to conduct a simplistic set of activities meant to plan and implement computing resources. Any discussion on applying the concepts of storage networking must touch on key workload definition concepts that are part and parcel to an effective capacity plan and to the possible rejuvenation of a more effective capacity planning model. Part V orients the reader towards applying storage networking concepts by way of an initial discussion and overview of workload definition and planning. The following chapters contained herein focus on identifying existing and anticipated workloads where storage networking solutions can be implemented for maximum value and productivity to the data center. SAN and NAS configurations will be addressed in analyzing the particular workloads of each.

We have explored and used the term workload many times already, and in Part V, it remains important to our discussion. Chapter 17 provides a detailed examination of workload definitions, identifications, and planning. Workload guidelines will be placed into a real-life context that characterizes both business applications and support activities, as well as the maintenance of capacity strategies as they relate to storage networking. Many of the activities that you manage on a daily basis may be enhanced or hurt by the new technologies involved in SAN and NAS solutions. In an effort to provide as much relative information as possible, generic descriptions and examples are used in Chapters 18 and 19 to cover several specific workload characterizations.

Part V addresses the usage of SAN and NAS as an integrated solution. Chapter 20, for instance, explores situations based on current IT experiences and also looks at future uses of integrated SAN and NAS solutions. This becomes a complex exercise when multiple workload scenarios are cast within the same environment. The ability of microkernels to communicate effectively and for storage caching mechanisms to operate in tandem becomes a challenge when considering the future of heterogeneous storage operation. This is compounded by the ability to maneuver within a distributed file system that incorporates optimum support for both traditional workloads yet sustains the proprietary nature and operation of database technologies. The current evolution of common application services through enhancement of web-enabled solutions will incorporate all the attributes of OLTP, batch, and data- intensive I/O, and will play a key role in these discussions.

Moving storage networking into the future requires an analysis of integrated storage networking solutions, using both FC- and IP-based connectivity schemes. Because of this, Chapter 20 addresses key future technologies, showing how they may affect IT storage networking solutions. Such technologies and standards include iSCSI, the Internet SCSI standard that allows SCSI commands to be encapsulated and transmitted through an IP network. In addition, we address the slowly evolving InfiniBand standard and infrastructure advancements that will result. Finally, well look at some of the internal bus connectivity standards and strategies that could have a dramatic effect on the data storage hierarchy as we know it today (see Chapter 5) exploring products such as AMDs Hyper-Transport, the industry standard VI transport mechanisms, and Intels next bus technology, Rapid-IO.