TieredStorage

Contents

• 1 What is tiered storage?

• 2 How does tiered storage work?

• 3 Why tiered storage?

• 4 When tiered storage is justified

What is tiered storage?

• Tiered storage is a system of assigning applications to different types of storage media based on application requirements. Factors considered in the allocation of storage type include the level of protection needed, performance requirements, speed of recovery, and many other considerations. Since assigning application data to specific media may be complex, some vendors provide software for automatically managing the process.

Figure 1 below helps explain where different types of storage are used in a typical data center, and the strategic fit of a corresponding storage approach. On the vertical axis is the importance of data availability and performance. Storage in tiers 1 & 2 support the most important applications to the organization, and tiers 2 & 4 support less mission critical applications. On the horizontal axis is the type of application that uses the data (operational or reference). In general there is a trend towards more data being mission critical, and a higher growth rate of reference data particularly driven by increased governance and regulatory requirements that mandate keeping data for longer periods of time.

The mission critical operational applications are in the top left hand corner. These are often database transactional systems, with a lot of reusable data suitable for caching, with high performance and recoverability requirements. Larger servers are driving the applications in this segment. The bottom right hand corner is often characterized by a large number of smaller servers and greater emphasis on connectivity.

As the amount of information grows, the challenge for the storage management staff is how to manage and optimize the different types of storage.

The importance of Figure 1 from a storage management perspective is a single storage solution is almost never appropriate for the entire data center. Having all storage with high functionality for example is not necessary or cost effective. Typically, one size does not fit all in the data center.

How does tiered storage work?

Conceptually, tiered storage is often depicted as a pyramid. In practice tiered storage involves software, hardware, processes and tools that facilitate the management of different classes of storage services across a wide range of workloads and applications. Key tiered software components include classification tools, monitoring and chargeback mechanisms, high performance data movers, hierarchical storage managers, access control facilities and resource managers which invoke or interact with a variety of storage services such as copy, backup and encryption.

Typically, storage tiers have grown as a function of individual application requirements and storage managers have been forced to develop enterprise architectures on the existing infrastructure in place. This has led to the need for a variety of storage management software techniques that rely on classifying data, data placement, data movement, backup and recovery and other management processes implemented across the device spectrum.

Why tiered storage?

There are three main strategic choices in managing a diverse and complex set of storage, which are:-

1. Mix, Match and Manage

This is the most common approach taken in data centers. The best fit of storage is applied to each appropriate application and the storage staff manages a diverse set of resources.

1. One storage system for all storage

Assuming high-end storage is used, this approach provides the lowest cost of administration and highest service levels to the user. The disadvantage is capital budget “sticker shock.” Alternatively, using low-end storage for all creates the reverse scenario – lower capital outlays, much higher overall costs of managing the storage.

1. A tiered storage software storage management overlay

Storage management software (either Server based or storage controller based) can provide an overall umbrella of functionality to all storage. For example, storage virtualization functionality can be provided by this type of software independent of the storage devices. The advantage of this approach is minimized storage costs, and common storage management processes and procedures across all types of storage. The disadvantages of this approach is the cost of the software, the potential tie-in to a vendors software, and the costs and risks of implementation.

Tiered Storage has the potential to significantly reduce the costs of storage and storage management functionality in environments when the cost of implementation and servicing is justified.

When tiered storage is justified

As stated above, a large number of data centers currently achieve a type of tiered storage environment by providing pools of different types of storage. The disadvantage of this approach is that different storage management software is usually required for each pool. Once an application has been assigned to a pool, it is difficult and expensive to migrate the application to another storage pool. There are also inherent costs in managing multiple different storage pools.

Tiered storage implementations are easiest to justify when:

1. There is significant change in the applications being delivered to the business

2. When there is a diverse set of applications providing services to many different parts of the organization

3. When the number of storage staff and the amount of storage are significant (unlikely to be achieved in small business data centers)

4. When there is good communication between the organization and IT on being able to define the business benefits of matching storage to the application

5. When IT has a good track record of deliving business value through its IT infrastructure.

Software and services to implement Tiered Storage are still maturing, and there are a wide range of potential solutions in the market place. Tiered storage consultants are in general wed to specific solutions that they know and have implemented. Practices that analyse the strategic choices of tiered storage solutions have yet to emerge.

In practice, this means that the majority of tiered storage software and services are being provided by storage vendors such as EMC, Hitachi, HP, IBM, NetApp, Sun and many other smaller vendors.

--------------X--------------

Implementing tiered storage

Many organizations are experiencing rapid growth in the amount of information held on computer storage and storage costs are escalating.

Traditional methods of managing storage use either a “one size fits all” strategy, or a "Mix, Match and Manage" approach introducing many different storage pools. The challenge of the first technique is high equipment cost, whereas the latter often brings higher labor costs to manage the multiple storage pools all with different software, services and procedures.

Today's storage technologies span a spectrum from high density, low cost, low performance storage at one end to increasingly sophisticated data management functionality at the other. Creating storage tiers with more granularity with a single management approach begins to address the problems that one size does not fit all and storage pools are hard to manage.

Tiered storage capability

Tiered storage is the process of provisioning different types of storage media based on application requirements. Storage tier choice may be based on performance, cost, protection needed, speed of recovery, and many other considerations. Ideally, a common set of storage services and procedures can be applied as required to each tier, matching application requirements with storage tier characteristics. A more complete description can be found on the WikiBon entry for Tiered storage.

Tiered storage has the potential to reduce the costs of storage and storage management functionality for some applications and improve the performance, availability, and recovery capabilities of others. The primary initial costs associated with a tiered storage implementation include software to manage the environment, implementation expenses, and possibly write-down costs for hardware that will be retired.

Specific operational goals of implementing tiered storage

The likely investment required to implement tiered storage is between $0.5 million to $0.7 million, with an elapsed time of between 6 to 9 months. [Note: These figures assume a Standard WikiBon business model organization with $1B in revenue with 4,000 employees and an IT budget of $40M per year. The scenario assumes that 40 terabytes are installed, with a mixture of high performance SAN, mid-range storage solutions, some direct attached storage, and some NAS]. A successful tiered storage implementation will:

• Allow consumers of storage to optimize the functionality, performance, and cost profile for storage that best fit the business need for each application by providing different tiers of storage.

• Ideally, provide a common set of storage management tools, processes and procedures and services for all the storage tiers, including the ability to dynamically migrate storage to the optimal tier.

The major expected reductions in IT cost will normally include:

• Fifteen percent (15%) reduction in cost of equipment and connectivity (~$200K/year).

• Twenty percent (20%) reduction in cost of storage administration (~$150K/year).

Case studies indicate the major expected business productivity improvements (so-called intangibles) will yield an estimated gain that equates to ~.06% of revenue and normally be derived from the following:

• Business benefits of accelerating time to implement new projects or speed changes (~$200k/year).

• Business value of improving security measured in mitigated losses (~$100k/year).

• Economic benefit of higher performance on user productivity (~$75k/year).

• Business value of improved productivity from reduced application downtime (~$25k/year).

In addition to the investment in tiered storage, other potential impacts on an organization include:

• Loss on books for storage equipment and software retired early.

• The business impact of delaying some projects because of resources dedicated to storage.

An analysis of the need for tiered storage is likely to find that a business case including both IT and business benefits will be good (ROI and IRR>150%), while a business case with only IT benefits is likely to be deemed acceptable (ROI & IRR ~100%). The business cases will be improved if organizations align spending on storage services with the additional business value these services generate in the supported applications.

Risks of implementing tiered storage

Tiered storage is a medium risk initiative. In general, tiered storage technologies, software, processes and procedures are still immature. The industry has yet to agree a common definition of tiered storage. In particular the current storage management software that monitors service level agreements at the application level can best be described as inadequate. There are four major risks to a tiered storage initiative:

1. Incorrectly defining the the number of tiers can increase costs.

2. Incorrectly designing the service level capabilities for each tier can negatively impact cost and business value.

3. Lack of tools to monitor and report storage on an application basis will negate the benefits of tiered storage.

4. Migrating to a common set of storage services risks compromising service levels for certain applications in the near term.

The tiered storage initiative

The tiered storage will be implemented when the tiered storage infrastructure has been designed, built, tested, implemented and successfully handed over to operations so that it can run as specified without external support.

Expectations (out-of-scope)

The following factors that are not within the scope of the tiered storage initiative are vital to ensure a successful tiered storage outcome:

• Application service level agreements are in place.

• A charge-back mechanism is in place that can be changed to charge user departments according to storage used and storage tier.

• Storage virtualization has been planned and/or implemented.

• The IT organization will support consolidating the storage management services and procedures.

• The primary storage vendors for the storage pools analyzed have solid tiered storage offerings.

Analyze phase

A fully detailed blueprint and step-by-step project outline for the analysis phase can be found at Analyzing tiered storage requirements.

Acceptance test considerations

The analyze phase will be completed when the initial business case has been accepted by the sponsor, and agreement has been reached to proceed to the design phase or kill the project.

Key analysis milestones

This phase should take about 4-6 weeks and 14-20 person days of effort.

1. An effective sponsor of the initiative is identified

   • It is important that the sponsor can resolve any organizational issues

2. Data collected

   • Determine the actual storage service level delivered for each of the major application groups

   • Determine the total costs of the storage, including complete staff costs, storage management costs, storage network costs, connectivity costs and (where applicable) telecommunication costs

   • Determine where there is either over provisioning or under-provisioning of storage services by application

   • Calculate the business and IT impact of that over/under-provisioning by application

3. An initial tiered storage design built:

   • Construct a basic tiered storage design and cost structure (one-time and ongoing).

   • Rules of thumb are that the number of tiers should be between 4 and 6, and is very unlikely to be outside the range of 2-8. The cost/GB of each tier up should be ~50%-100% more expensive.

4. Business case constructed:

   • Analyst constructs business case / cost benefit analysis detail.

5. Initial Design and business case accepted by sponsor and any other stakeholders necessary

Design phase

Individuals interested in a detailed discussion of tiered storage design should check out Designing tiered storage.

Acceptance test considerations

The design phase will be completed when the design has been accepted by the sponsor and agreed to by the key application user groups, and agreement has been reached to proceed to the deploy phase or kill the initiative.

Key design milestones

This phase should take about 5-7 weeks and about 30 person days of effort.

1. The number of storage tiers required established

   • There is a trade-off between the number of storage tiers provided, and the overhead of managing and migrating data between the different tiers. In practice, the number of tiers can be determined by:

     • Taking the key applications that support the business (e.g., CRM, Email, Finance, Reporting, etc.). Create a table with the characteristics required for each of the major key applications. The figures in the table are not what is currently provided, but a judgment on what the most appropriate storage service level is to support the efficiency of the users.

     • Group the applications into similar tiers, reducing the number of tiers. Usually, the number of tiers will be between three and six. In smaller installations it may be as low as 2, but should not be above 10 under almost any circumstances.

2. The storage network implications analyzed and the costs/savings of additional/reduced ports, switches, and other network costs computed

3. A storage cost/Gigabyte calculated for each tier, with a rule of thumb that each successive tier should be between 50%-100% more expensive

4. The storage tier(s) appropriate for the major applications agreed with the user departments, together with the growth rates expected, and expected storage budgets and storage service levels set for each major application. This agreement process has been pushed down to the rest of the application groups

Deploy phase

Acceptance test considerations

The deploy phase will be completed when tiered storage is built, tested, and brought into service. The agreed service levels for each tier should have been met for at least 3 months, and the operations group able to operate independently of the design and implementation teams.

Key deployment milestones

This phase should take about 4-6 months and cost between $0.4million & $0.6million.

1. Tiered storage built

   • A tiered storage system RFP should have been defined and issued, including (optimally) virtualization capabilities, common storage management capabilities, and the ability to migrate from the existing infrastructure to the tiers and between tiers without interruption

   • Installation of storage hardware and storage management functionality

   • Update and creation of new process and procedures, with full documentation

2. Tiered storage tested

   • Testing of equipment, software, and procedures on historical data

   • Testing of tiers on some non-mission critical live applications

   • Testing of procedures for migration, backup, recovery, and disaster recovery

   • Tiers supporting key applications stress tested

3. Migration & Cut-over to tiered storage completed

   • Phased migration cut-over to tiered storage

   • Extensive monitoring of performance, reliability, & storage efficiency metrics (storage utilization, administration efficiency, planned vs. actual allocation of storage to tiers)

4. Tiered storage initiative wrapped up

   • Procedures set up for monitoring expected/actual achievement on key storage metrics

   • Procedures set up for adding additional storage & storage functionality

   • Final review of documentation

   • All project staff released and full hand-over to storage operations

Tiered storage summary

Tiered storage is a technology that makes intrinsic sense. Most organizations are likely to determine that a tiered storage implementation should be in their strategic plans. When organizations have the storage architecture foundations in place, such as centralization, virtualization, and a commitment to a single management suite, early implementation should be considered. If the foundations are immature, tiered storage may wait or be implemented in an a part of the organization ready to exploit the solution.

External Links:

-----------------

1. http://wikibon.org/wiki/v/Tiered_storage

2. http://wikibon.org/wiki/v/Implementing_tiered_storage

3. IBM adds Easy Tier to DS8700 (Auto data movement to SSD)

http://silvertonconsulting.com/cms1/tag/easy-tier/