Consolidating Storage
Today, there is as much interest in storage consolidation as there is in server consolidation. As we said earlier, every new server you deploy results in more storage. In many cases, the cost of storage for a server exceeds the cost of the server, and although the server may not grow very much, the amount of storage required for an application will grow indefinitely.
When you look at application sets, you find a lot of data replication because multiple applications look at the same data. For example, if you have an OLTP server, a batch server, a data warehouse server, and four data mart servers, you may have seven copies of a particular dataset such as a customer master file. While, in theory, all seven copies are identical, because different developers probably created their applications, it is likely that there will be differences in each copy of the data. This situation has the potential to create a situation where reports run on the "same" data, yet yield different results.
As with servers, the goals in storage consolidation are to reduce complexity, increase utilization, and reduce TCO. These goals can be achieved through a variety of consolidation techniques. The ultimate goal is data sharing among applications. Unfortunately, this often requires redesigning and redeveloping applications, so it is a long-term goal, at least for backward consolidation. In a forward consolidation, data sharing should absolutely be a goal.
Other benefits of storage consolidation include:
Easier backup and recovery
Increased availability
Improved scalability
Storage resource pooling
When undertaking a storage consolidation effort, it is critical that you understand how disk space is utilized. While most of our customers can't tell us exactly what their current disk space utilization rates are, many companies we have surveyed estimate a rate of 40 percent, or less. For companies that have tried to accurately assess disk utilization, complexity often hinders their efforts. For example, one company we visited started counting storage, but stopped at 110 terabytes after realizing that they just couldn't count it all. When we evaluated their disk utilization, we found they were utilizing only 20 percent of their available storage.
There are several types of storage consolidation available today. The following sections describe the three most common types we see.
Consolidating Servers and Their Associated Storage
With every server consolidation, there is an accompanying storage consolidation. As you move multiple applications to a single instance of the OS, you must also move their storage, as shown in the following graphic. In theory, once you have moved the storage, data will be available to any of the applications. This is the most primitive form of storage consolidation.
FIGURE 0-2 Consolidating Direct-Attached Storage
Connecting Heterogeneous Environments to a Single Storage Component
Many of our customers have heterogeneous IT environments. They run servers and OSes from many vendors, and they want to access data from a variety of different servers. With direct-attached storage, this was difficult to do. Now, with products like the Sun StorEdge™ 9900 storage array, it is possible to connect the Solaris™ Operating Environment servers, other UNIX servers, Microsoft Windows NT servers, and mainframe servers to the same storage array. The following graphic demonstrates this capability.
FIGURE 0-3 Heterogeneous Environments to Single Storage
This is a very popular type of storage consolidation, especially where there are multiple mission-critical applications running on servers from a variety of vendors.
Consolidating With Storage Area Networks
Storage area networks (SAN) have been the hottest trend in storage architectures for the last few years. As a technology, the SAN is now mature enough that it can be implemented using standards and standard configurations. As shown in the following graphic, SAN technology inserts a network, or fabric, of switching devices between servers and storage that enable any server or application to access any storage connected to it. The fabric can then be configured to allow various servers to access various storage.
FIGURE 0-4 Storage Area Network (SAN) Configuration
Another hot storage technology is network attached storage (NAS). This technology allows servers and clients to utilize storage directly over the network using common networking protocols like network file system (NFS) and Server Message Block (SMB). Although not used greatly in many server and application consolidations within the data center, it is used extensively in file and print consolidations at the department and workgroup level. As the technology matures, expect it to work itself into data center consolidations.