Storage Array Migration Costs

Premise

Wikibon research compares the 4-year total cost of acquisition, maintenance and migration costs of a typical $300,000 traditional storage array with a Server SAN approach in a mixed workload environment. We also include data for a complex Oracle environment. The Server SAN approach is between 45% to 48% lower in total 4-year costs.

Executive Summary – The Hidden Storage Migration Costs

Storage migrations are much more costly than most people realize because of hidden costs. The Summary Table below shows the components of an array migration for a mixed workload and a more complex Oracle environment. The data shows total four-year costs for two types of array architectures, a traditional storage array and a modern architecture represented by Server SAN.

Based on industry surveys of Wikibon community practitioners, the components of the array migration include:

• The need to extend the maintenance on an older array to adequately prepare and plan for the migration;
• The purchase of the new array which must be done prior to the migration;
• The migration procedure itself, including the planning, training and procedure update work;
• Professional services associated with critical external skills the migration.

Based on our data Wikibon estimates the average four-year cost of a traditional $300,000 array migration is $191,000 for a mixed workload, more than 60% of the initial acquisition costs. Note that when migration is to a Server SAN class of product, this figure drops to $110,500 or 36%. However because capacity for the modern storage solution can be acquired in more granular increments, customers will save on the need to acquire new hardware early (because they can buy in chunks), effectively making the net migration costs negligible for a mixed workload.

Migration costs increase substantially in an Oracle workload because of additional complexity.

Evaluating Migration Costs in the Context of Total Cost of Ownership

Figure 1 shows the 4-year cost comparison between a traditional storage array and a ServerSAN approach. The cost comparisons include acquisition costs, maintenance, operational staff, consulting and migration costs. Two workload environments are analyzed, a typical mixed workload environment and a complex Oracle workload environment.

The results in Figure 1 show a very significant reduction in both CAPEX and OPEX 4-year costs when Server SAN is deployed. In a typical mixed workload environment, Server SAN is 48% lower cost. In a complex Oracle Database environment, Server SAN is 45% lower cost.

One fundamental reason for the lower migration costs for Server SAN over traditional storage arrays is the movement of software from a proprietary storage controller hardware to software executed in a general purpose server. This creates independence between the server power and storage capacity required for storage.

Methodology for Migration Costs

Wikibon has estimated storage array migration costs several times. In 2014 Wikibon set out the basic framework of estimating the cost of migration from one traditional storage array to another. In 2016 Wikibon looked at the cost of Migration in a complex Oracle environment.

Our latest estimate for 2017 is shown in Figure 1 in the executive summary above. The assumptions are shown in Table 2 and Table 3 in the Footnotes below.

The three major components of migration are

1. The Project Cost of a Migration:
   • Wikibon has estimated the cost of a storage migration project to include new storage costs, maintenance, operational costs, skills upgrade costs, storage procedure costs, and consultancy costs. The details are shown in Table 2 and Table 2 in the Footnotes below.
   • One of the major complicating factors of storage migration is planning and testing for an outage before and/or after the migration is completed. As a result, migration cut-over is usually planned on an application-by-application basis, involving the application owners. Application down-time is increasingly being resisted, as user expectations of 24 x 7 application availability increase. The planning by application of migration is typically spread over a eight month period for a typical mixed workload environment. For the complex Oracle workload environment, the assumption is that the total migration time is 16 months, with a requirement to keep the old storage array for 12 months.
2. The Cost of Buying the New Storage Array or Server SAN Early:
   • The new storage array has to be brought in about five months early to accommodate the planning and elapsed time for data migrations. Table 2 and Table 3 in the Footnotes below shows the assumptions for different workload environments.
3. The Cost of Keeping the Old Storage Array Longer
   • After the final migration, the old storage array has to be kept in place in case of problems requiring access to the original data, and for the application to be able to “fail-back” to the original storage if necessary. Table 2 and Table 3  in the Footnotes below shows the assumption for different workload environments.

In general, the cost and elapsed time of storage array migrations is reducing slightly, as the new storage arrays and Server SAN architectures have much higher levels of flash storage capacity. Flash speeds up migration and lowers the cost of operational staff to manage it. Federated storage architectures such as HP’s Peer Motion and NetApps Clustered Data ONTAP can significantly ease migration between the arrays in a federation. Putting all the storage under a virtualizing software umbrella (SVC from IBM, VPLEX from EMC, VSP from Hitachi, etc.) eases migration but brings other overheads. The fundamental architectural problem is that each traditional storage array has the storage management built-in to the hardware architecture.

Action Item: Storage migration continues to be a major cost element for traditional array-based storage in 2017. Senior IT executives should ensure the cost of migration is built-in to the business case for new storage arrays and should strong consider architectures like Server SAN which can significantly reduce the elapsed time and cost of provisioning and retiring storage.

Footnotes: Storage Upgrade and Migration Costs