Tiered storage cuts costs, improves business alignment.

Business managers want tools that automate, monitor, and audit the storage and migration of data among platforms. Using new, lower-cost storage technologies and better aligning service levels with the value of the data, a tiered approach can reduce total cost of storage by upwards of 45% while improving the strategic alignment of storage investment with business needs.

Why Tiered Storage?

One reason businesses need a tiered approach to storage is the sheer volume of information they must store.

Even through the economic downturn of the last few years, data storage requirements have continued to grow by 70-80% a year for many organizations. This has been driven by the increased use of information systems to increase the efficiency of logistics, supply chain and distribution systems, by the storage needs of e-commerce functions such as online sales, marketing and collaboration with business partners; and the needs of specific applications such as enterprise resource planning (ERP) and e-mail. Regulations that require the long-term storage of critical data are also driving storage growth in vertical markets such as financial services and health care.

A second factor is the growth in unstructured data such as documents and images. Unlike information stored in the rows and columns of a traditional database, unstructured data may include anything from a mortgage application to a patient x-ray. It represents the fastest-growing type of data, increasing at rates between 120% and 200% a year, driven by the increasing use of work-flow applications in financial services, which relies heavily on documents, and health insurance, which relies heavily on images.

Unstructured data drives up the need for tiered storage because its value tends to fall more quickly over time than does the value of structured data. For example, an e-mail may be critical the first day it is sent, but after a week may never be opened or needed again. This means unstructured data can be archived sooner on slower, less expensive media than structured data, provided the organization has a tiered storage infrastructure which allows such migration.

Another driver is stagnant or even shrinking IT storage budgets, a situation that may not improve even as the economy improves. Having discovered they can "do more with less," management will in many cases be reluctant to increase spending for storage unless the IT staff can prove a compelling return on investment. Smaller IT staffs also mean companies have fewer skilled people to evaluate, choose, and integrate multiple technologies from multiple vendors, as well as fewer staff to perform ongoing storage management. Packaged solutions to address common pain points like e-mail and database archiving as part of a tiered storage architecture can help reduce complexity while delivering solid business benefits.

At the same time, business managers have become used to getting higher performance, reliability, and availability year over year even if they are paying the same or less for each megabyte of storage. Tiered storage can help IT organizations deliver such improvements while reducing the long-term total cost of ownership of storage.

In some industries, new regulations requiring the long-term retention, proof of integrity and timely recoverability of data are driving the need for tiered storage architectures. Among the well-known examples, the Sarbanes-Oxley Act requires auditors to keep records of financial audits for seven years, while HIPAA requires health care providers to keep patient records for two years after a patient's death, and to guarantee access to such records in case of disaster. By providing different storage service levels and recovery times for different types of data, tiered storage can help organizations meet such regulatory requirements without breaking the bank.

The final trend driving tiered storage is new technologies that enable a tiered-storage strategy, such as ATA-based disk arrays that offer disk-like performance for tape-like costs. New fixed-content storage platforms offer new cost-effective options for long-term data storage, while new storage management software provides new capabilities for managing and moving storage as needed across different storage devices.

What is Tiered Storage?

Tiered storage refers to different combinations of storage hardware and software designed to provide specific levels of performance and functions at various price levels.

It provides up-front savings because customers purchase only the storage capabilities they need for their parti-cular application requirements instead of overbuying to meet "worst case" or "peak demand" needs. It also provides ongoing cost savings because customers now have the tools and processes to more effectively manage the storage they have, and to ensure their current storage resources are aligned with business requirements before purchasing more.

Implementing a tiered-storage strategy also provides a better dialogue between IT and the business about the importance of various application information to the business, and about the price/performance tradeoffs available with various storage technologies.

Tiers are defined, of course, by cost as measured by purchase price per megabyte and by ongoing management costs. They are also defined by other quantifiable measures such as the Recovery Point Objective (RPO), which is a measure of what data and how much data the business needs to be able to recover if a system failure or disaster occurs, as well as the Recovery Time Objective (RTO), which refers to how quickly the data must be recovered.

Tiers can also be defined by the presence or absence of functions such as remote copy, mirroring, replication, and auditability. In some industries, for example, auditors or regulators may require time stamps, checksums or other proofs that long-term archived data has not been altered, or a record of who made any changes to it and when.

Implementing Tiered Storage

As with many strategic IT initiatives, the move to tiered storage begins with a dialogue between the storage architects and the business managers. The aim of this conversation is to classify the organization's applications by their importance to the business, since storage is important only because the data it stores is used by applications.

Such a classification will take into account factors such as the applications' effects on cash flow, profits, customer satisfaction, as well as the effect on revenue, profits, or even shareholder price if the application is not available. This classification may include measures such as RTO and RPO, and how quickly the data associated with the application may become more or less critical to the company. A company operating in a fast-changing regulatory environment, for example, will want the flexibility to move certain data to storage systems that provide faster response times if regulators require it.

The business managers and IT staff then define the tiers of storage (by price, performance and functions) needed to support these different levels of applications. A simple architecture might consist of three tiers called high-availability, mid-tier, and archive. The high-availability tier might include direct-attached or Fibre Channel SAN storage with data protection features; the mid-tier might be made up of lower-cost, but still relatively fast ATA based storage networks with fewer data protection features; and the archive tier might consist of still lower-priced and lower-performing tape.

In other cases, customers have defined between 6 and 10 tiers of storage by creating tiers based on very specific application needs. A financial exchange, for example, might have both a "highest-availability" tier of storage for the trading application used during business hours and a "high-availability" tier (which might be slightly less reliable and somewhat slower) that is perfectly adequate for overnight reconciliation and reporting applications. Some organizations might require a separate storage tier for testing and application development. In some cases, even small differences in performance, reliability, or recovery time can result in major cost implications, with total annual costs varying by a factor of 100 to 1 depending on the tier selected.

The final step is to make these tiers visible to the business, which means going back to the application categories and explaining which storage tiers meet the needs of which applications. Here, the IT staff presents the business managers with a range of options and associated prices for levels of uptime, recovery time, and application response time.

Any IT manager for whom storage is a strategic corporate concern should already be working with his business partners to classify the organization's applications. One good way to prove the tiered storage concept is to start with applications that are posing clear storage problems.

One example of such "low-hanging fruit" is e-mail, where large but never-used attachments such as presentations can be moved to less-expensive off-line storage to provide an immediate return on investment. Another example is ERP databases that have been in use for years but never examined with an eye to moving old or irrelevant data onto less-expensive storage platforms.

Tiering vs. Other Storage Architectures

Storage veterans will recognize similarities between tiering and the hierarchical storage management (HSM) already practiced in many data centers, where data is moved from fast disk to tape as its criticality to the business decreases. However, the new business imperatives make tiering far more strategic than HSM, while new storage technologies make tiering far more powerful.

HSM typically dealt only with files but not databases or unstructured data, which is the fastest-growing area for storage demand. A typical HSM environment usually had only two tiers: fast, expensive disk or slow inexpensive tape. These options are too limited to satisfy today's cost-constrained business manager, and are dwarfed by the tiering options available with today's low-priced ATA disk technology and advanced storage management software.

Many business managers have adopted a simple form of tiering for years through the service level agreements (SLAs) requiring IT staffs to provide certain levels of application performance to end users. However, most SLAs offer too narrow a choice of service levels to reflect the actual needs of the business, and don't offer business managers the full range of tradeoffs and options available to them. For example, most SLAs deal only with application response time rather than today's business-critical issues such as the speed of data recovery or whether the application is using storage as efficiently as possible.

Making the move to tiered storage is not a one-time process or the reaction to a temporary problem. It is instead a long-term, structural change in how customers think about, choose and implement storage.

Tiered storage saves money up-front and paves the way for not only continued efficiency, but also improves alignment between business and IT. It is a critical essential element of competitiveness. If you aren't already well on the way, one of your competitors is.

www.emc.com

Chuck Hollis is the vice president of Platform Marketing at EMC Corporation (Hopkington, MA)