Application-Centric Management

Application-Centric Management

BMC Software was among the first to note the deficit of the SRM approach to storage management and, in 2000, began pursuing an alternative strategy of application-centric storage management. ^[7] The idea was to segregate the horizontal view of the storage infrastructure on an application-by-application basis and to tie all storage management activity to application performance metrics (see Figure 8-5).

Figure 8-5. Application-centric storage management.

While this strategy would not have, by itself, provided a solution for storage utilization efficiency, application-centric management suggested an effective replacement for the intelligent automatic provisioning that SANs were supposed to deliver, but never did.

BMC began to complement its Patrol Storage Manager product with a set of tools that would enable manual storage provisioning ”with application performance providing a means to validate task completion. So practical and successful was this approach that other "800- pound gorillas" in the market, including Veritas and EMC, began chanting the BMC mantra, and IBM was both using BMC Software in its own facilities while proposing it in large storage deals in place of its own SRM tools from Tivoli.

BMC preached application-centrism for four years until their sudden and unexpected departure from the market in early 2003. Other vendors have sought to pick up the mantra, but the results have proven the truth of the old saw that there is a big difference between "talking the talk" and "walking the walk."

Many products can, after a fashion, emulate the application-centric presentation of the data storage infrastructure: a baseline functionality requirement for an application-centric management approach, as shown in Figure 8-6. However, few offered the tools that BMC provided to enable the administrator to intervene based on whatever hot spots or choke points were discovered .

Figure 8-6. Hierarchy of functionality in application-centric storage management.

For example, if a database index and data set are written to the same storage platter, application performance may actually suffer as the result of contention for read/write heads. An effective application-centric management scheme provides both the means for visualizing this condition and also the tools to move stored data around between and among platters to aid in fixing such hot spots or choke points.

Going further up the hierarchy of application-centric storage management, such a product must also capture data on processes or procedures frequently undertaken to resolve common issues and provide an engine for automating responses to such conditions whenever they arise. BMC, through its relationship with Invio, was well on its way to achieving such functionality. As of this writing, none of the company's successors were as far along on this path .

The pinnacle of the application-centric storage management functional hierarchy consists of the ability to automatically allocate storage to an application and to manage that storage through the tools and policies located lower in the function stack. Such an approach requires a knowledge base containing typical data storage requirements for popular application software, tunable by customers to meet their specific needs, that would interface with a second knowledge base containing the specific details of the storage platforms deployed in the customer's shop. In operation, new applications would be provisioned with storage automatically, then reprovisioned per policy whenever resources neared a given threshold.

Figure 8-7 provides a model for such a system that is being pursued by several " name brand" SRM vendors today. The "storage management server" uses a virtualization (that is, LUN aggregation) layer to track storage resources and provide dynamic scaling in response to an automated response engine. SRM tools and an event monitoring function help to keep the storage platform in proper operating condition. Atop the stack is an ACSM-like policy engine that provides the functionality depicted in the application-centric storage management model.

Figure 8-7. High level architecture of an ACSM-based storage management server.

Even had BMC's efforts been fully realized, it is doubtful that improved storage capacity utilization efficiency would have resulted from such a strategy. Application-centric storage management concerned itself pragmatically with the alleviation of the storage provisioning pain confronting IT professionals and less with optimizing capacity allocation and data migration within storage management infrastructure. It was very much a descendent of SMS from the IBM mainframe environment.

However, utilization efficiency requires the consideration of another dimension of storage management: data access patterns. It is difficult to see how an application-centric storage management approach could effectively improve storage utilization since it does not consider how data itself is used.