Data Center Measurement, Metrics & Capacity Planning

Part six in a series. Greg Schulz is the founder of StorageIO and the author of The Green and Virtual Data Center.

Good decision making requires timely and insightful information. Key to making informed decisions along with implementing strategies is having insight into IT resource usage and the services being delivered. Information about what IT resources exist, how they are being used, and the level of service being delivered is essential to identifying areas of improvement to boost productivity, raise efficiency, and reduce costs. It is important to identify metrics and techniques that enable timely and informed decisions to boost efficiency in the most applicable manner to meet IT service requirements.

The overall health and status of the equipment, the steady supply of raw materials, energy or power supply, and the quality of service need to be constantly measured for timely management decisions to be made. For example, in an electrical power plant, control rooms' or operations' nerve centers closely monitor different aspects of the facility and its resources, including fuel supply and cost, production schedules and forecasts, boilers, exhaust air scrubbers, water chillers and associated pumps, turbines, and transmission status.

A common focus, particularly for environments looking to use virtualization for server consolidation, is server utilization. Server utilization does provide a partial picture; however, it is important to look also at performance and availability for additional insight into how a server is running. For example, a server may operate at a given low utilization rate to meet application service-level response time or performance requirements. For networks, including switches, routers, bridges, gateways, and other specialized appliances, several metrics may be considered, including usage or utilization; performance in terms of number of frames, packets, IOPS, or bandwidth per second; and latency, errors, or queues indicating network congestion or bottlenecks.

From a storage standpoint, metrics should reflect performance in terms of IOPS, bandwidth, and latency for various types of workloads. Availability metrics reflect how much time, or what percent of time, the storage is available or ready for use. Capacity metrics reflect how much or what percentage of a storage system is being used. Energy metrics can be combined with performance, availability, and capacity metrics to determine energy efficiency. Storage system capacity metrics should also reflect various native storage capacities in terms of raw, unconfigured, and configured capacity. Storage granularity can be assessed on a total usable storage system (block, file, and object based and content accessible storage-cas) disk or tape basis or on a media enclosure basis -- for example, disk shelves enclosure or individual device (spindle) basis. Another dimension is the footprint of the storage solution, such as the floor space and rack space and that may include height, weight, width, depth, or number of floor tiles.Measuring IT resources across different types of resources, including multiple tiers, categories, types, functions, and cost (price bands) of servers, storage, and networking technologies, is not a trivial task. However, IT resource metrics can be addressed over time to address performance, availability, capacity, and energy to achieve a given level of work or service delivered under different conditions.

It is important to avoid trying to do too much with a single or limited metric that compares too many different facets of resource usage. For example, simply comparing all IT equipment from an inactive, idle perspective does not reflect productivity and energy efficiency for doing useful work. Likewise, not considering low-power modes ignores energy-saving opportunities during low-activity periods. Focusing only on storage or server utilization or capacity per given footprint does not tell how much useful work can be done in that footprint per unit of energy at a given cost and service delivery level.

Virtual data centers require physical resources to function efficiently and in a green or environmentally friendly manner. Thus it is vital to understand the value of resource performance, availability, capacity, and energy usage to deliver various IT services. Understanding the relationship between different resources and how they are used is important to gauge improvement and productivity as well as data center efficiency. For example, while the cost per raw terabyte may seem relatively inexpensive, the cost for I/O response time performance needs to be considered for active data.

Having enough resources to support business and application needs is essential to a resilient storage network. Without adequate storage and storage networking resources, availability and performance can be negatively impacted. Poor metrics and information can lead to poor decisions and management. Establish availability, performance, response time, and other objectives to gauge and measure performance of the end-to-end storage and storage networking infrastructure. Be practical, as it can be easy to get caught up in the details and lose sight of the bigger picture and objectives.

Greg Schulz is the founder of StorageIO, an IT industry research and consulting firm. He has worked as a programmer, systems administrator, disaster recovery consultant, and capacity planner for various IT organizations, and also has held positions with industry vendors. He is author of the new book The Green and Virtual Data Center (CRC) and of the SNIA-endorsed book Resilient Storage Networks (Elsevier).InformationWeek Analytics has published an independent analysis of green IT strategies. Download the report here (registration required).0