In January 2010 Wikibon discussed different storage approaches to supporting Microsoft Exchange. The conclusion was that for user seat configurations below about 800, direct access storage (DAS) was lower cost, while above 800 seats storage area network (SAN) configurations were cheaper. The key economic issue is that DAS is not as fungible an asset as SAN that can be shared across the application portfolio easily. As a result, for narrow application suites, DAS may be appropriate and less expensive but in general, for most organizations of any scale, SAN will be a more attractive economic alternative. As well, politics come into play as sometimes application groups want to control their own storage destiny.
In 2010 Microsoft played off of these dynamics and was strongly promoting DAS as the least expensive implementation. Microsoft was also promoting the use of its replication services as being equivalent to SAN replication technologies. The reason for this is shown in Figure 1, where the costs of storage hardware, software, and archiving services are equivalent to the difference between the on-premise Exchange and MS Exchange Online. If storage can be made less expensive by using DAS, Microsoft could reduce the cost competitiveness of on-premise Exchange against cloud alternatives and improve margins.
In January 2011, Microsoft took a more practical stance real-world, and included SAN configurations in its recommendations. Wikibon members are interested in whether additional functionality in storage systems over the last two years has changed the findings and whether a requirement for very high availability SLAs would change the recommendations.
The conclusions of the analysis are shown in Figure 2. The analysis shows that for “standard” configurations, the cost of SAN was lower than DAS for 500 seats and higher. For high availability (HA) configurations, SAN was always significantly lower in cost: the average cost of DAS configurations was about 39% higher than SAN configurations.
Wikibon concludes that the cost of DAS is about 11% higher for most standard availability configurations over 500 seats. When higher availability is required, the SAN storage arrays and appropriate software management tools provide enhanced capabilities for meeting much higher RPO and RTO levels. The cost of SAN becomes 39% higher than the DAS alternatives.
For Microsoft Exchange infrastructure with significant high availability requirements, Wikibon would recommend SAN as the default approach, which will result in significantly lower cost than DAS configurations. For standard availability requirements, SAN would also be recommended except for very small static environments (<400 seats where there is very little probability of future growth in Exchange users).
For Exchange installations, Wikibon would also recommend using array-based application consistent space-efficient snapshot technologies together with asynchronous replication of the snapshots to a remote site. Also recommended is management software to manage the snapshots and recovery of data from them, and monitoring availability performance against availability service level agreements.
Model Assumptions: ROA and NetApp Infrastructure
For our Exchange models, Wikibon uses a technique focused on measuring return on assets (ROA). It’s similar to ROI but differs in that it attempts to capture the cost metrics of installed infrastructure over a life cycle. We modeled DAS versus SAN and measured only costs, not other business value. Specifically for this analysis Wikibon assumed the use of NetApp snapshot hardware, together with the appropriate SnapManager software as the reference architecture. NetApp's value proposition to clients is largely based on a unified storage architecture that makes the same storage efficiency and management functionality available for all data, regardless of differences in the underlying hardware. The core to NetApp’s product strategy is the ONTAP proprietary operating system, built round the WAFL (write anywhere file layout) file system. The operating system started with support for NFS and has expanded into a unified storage system supporting CIFS, iSCSI and FC, creating that unified architecture.
NetApp's marketing thrust is to emphasize the efficiency of its systems relative to other traditional SAN approaches. Wikibon has confirmed the NetApp approach offers value and is differentiated in the market. As always, innovations have tradeoffs and a good discussion of WAFL's capabilities and tradeoffs can be found here on Wikibon. Generally users we've interviewed have indicated that they have not encountered widespread technical or performance issues with NetApp technologies.
Action Item: Availability is a key requirement for almost all e-mail systems, and is much easier to achieve using array-based snapshot technologies linked to Microsoft VSS services that ensure Exchange application consistency. Wikibon would strongly recommend CIOs, CTOs, and Microsoft Exchange managers to ensure that SAN snapshot-based availability technologies are used in preference to direct attachment of storage for all Microsoft Exchange Environments that need or may need to meet world class RPO and RTO service level objectives.Footnotes:
The High Availability Assumptions assumed the use of NetApp Snapshot Technology, including SnapVault, Protection Manager, Data Fabric Manager, Exchange Manager, Asynchronous Replication and High-speed data connection.