Common management of multi-vendor environments and agent-less management are two of the benefits of the Storage Management Initiative Specification (SMI-S).
By John Kelly
If you are in the data storage industry or IT infrastructure management, you have either heard of the Storage Management Initiative Specification (SMI-S) or will soon. For some storage professionals, SMI-S solves all management problems, while others are more skeptical. The truth is somewhere in the middle.
The continued explosion of data that needs to be stored and managed is not news. Also, the fact that the cost of storage management has far outstripped the cost of storage acquisition is well-documented and well-understood. The SMI-S storage management standard addresses these challenges.
SMI-S v1.0.2, which most storage products are based on today, provides the first step toward making multi-vendor storage environments simpler to implement and easier to manage. Products that conform to the SMI-S standard are delivering value today by enabling simpler configuration and allocation procedures, standardized controls for complex operations, and a foundation for policy management that will increase the efficiency of provisioning and fault management.
Moving forward, the consolidation of discovery, access, reporting, and control features into a single, standard architecture will establish a foundation for a broader set of management capabilities in the future. SMI-S v1.1, for instance, which vendors are currently testing their products against through the Storage Networking Industry Association Conformance Testing Program (SNIA-CTP), focuses on higher-level services such as information lifecycle management (ILM), copy services, security, and policy management. SMI-S 1.1 also introduces management of additional storage resource types such as tape libraries, NAS, and iSCSI, moving SMI-S beyond conventional Fibre Channel SAN management to delivering higher-level data services for all storage technologies.
More importantly, however, is the fact that nearly all leading storage resource management (SRM) vendors have implemented the SMI-S standard. Users and vendors are beginning to realize the following:
■ Storage administrator benefits: Normalized multi-vendor array, switch, and host management; agentless management of Windows and Unix infrastructure; and accelerated support of new devices and increased freedom of choice gained from having standard device interfaces; and
■ Storage vendor benefits: Lower cost of development, testing, and training; increased choice of solutions that work in end-user environments; and accelerated time to market.
Now let’s explore how storage management applications are leveraging SMI-S to reduce complexity and management costs today through normalized multi-vendor management and agent-less management of Windows and Unix Infrastructure.
Normalized multi-vendor management
Storage networks were introduced to ease issues with backup as well as improve access rates, availability, and the efficiency of managing distributed storage. This is accomplished by centralizing operations and allowing hosts running multiple operating systems to share consolidated storage.
Storage networks also introduced many new layers of abstraction and network resources to manage, thereby replacing the direct server-to-storage relationship. This problem was amplified as users were forced to master multiple device managers, fabric managers, performance tools, etc., resulting in the soaring cost of training and administration. A key objective of SMI-S is to reduce complexity and management costs through normalized management of multi-vendor arrays, SAN switches, and hosts in storage management applications.
In particular, storage management applications leverage the SMI-S definition of how a certain resource type (e.g., arrays, switches, and hosts) should be modeled in the user interface, thus offering administrators a common-or normalized-way in which to manage resources of the same type from multiple vendors.
For example, as shown in the figure, storage management applications based on the SMI-S model provide a common means to manage disk arrays from different vendors-in this case storage arrays from two vendors can be managed in the same manner. The same can be said for managing switches and hosts from multiple vendors. Having a common interface design not only allows administrators to become more efficient at managing resources and consequently service levels, but also reduces training and allows administrators to ramp-up faster.
IT organizations are constantly looking for ways to more-efficiently manage their server and storage infrastructures. Today, most management solutions require the deployment of host agents, which impact the performance of each managed server, raise memory requirements, and increase administrative overhead. However, SMI-S- based storage management solutions leverage Windows Management Instrumentation (WMI)-Microsoft’s implementation of the Common Information Model (CIM) standard-and Sun’s WBEM Server to deliver agent-less management of Windows and Solaris infrastructure and support strategic IT projects such as server consolidation and asset management.
Normalized array management
Similarly, The Open Group’s Pegasus project is an open-source implementation of the SNIA’s Data Management Task Force (DMTF) CIM and WBEM standards that is designed to run on most versions of Unix, Linux, and Windows, and most of the large Unix vendors are working on implementing CIM Object Managers (CIMOM) into their operating systems.
Microsoft’s WMI, for example, is configured as a default service during Windows server installation and represents data-center components in a common object model. By directly interfacing with these resources through WMI, SMI-S-based storage management applications provide extensive management of Windows-based infrastructures without having to install proprietary management clients, or agents. The box on the left shows the agent-less management features in SMI-S-based storage management applications, which can help administrators take the initial step of automating discovery and asset management.
For example, the screen capture in the figure depicts the capacity utilization information that can be culled from a Windows server through WMI. Capacity details (i.e., total capacity, volume name, partition type) are presented for File Allocation Table (FAT) and Windows NT File System (NTFS), giving administrators a central way to view and manage capacity utilization. Similar levels of detail are provided for tracking real-time, historic, and forecasted performance parameters.
WMI can also be extended by adding CIM extensions that piggyback on the existing WMI service already running on Windows servers. By contrast, non-SMI-S SRM solutions require that dedicated, proprietary agents be installed on Windows servers before any management features are available.
SMI-S-based products provide management savings for multi-vendor storage environments by enabling procurement of best-in-class products for any particular application and reducing agent proliferation across the enterprise. This, in turn, reduces the overhead and complexity of managing ever-larger amounts of storage per individual by providing the ability to manage all vendors’ storage assets from a single management application.
To learn more about the SNIA’s SMI-S standard, visit www.snia.org/smi.
John Kelly is vice chair of the SNIA Storage Management Forum and a storage software product marketing manager at Hewlett-Packard.
Agent-less management features delivered by SMI-S
■ Automatic discovery of all Windows servers
■ Detailed configuration information about every server, including vendor, model, operating system version, installed memory, and CPU
■ Logical storage volume information, including mount points, physical devices, drive types, and file system types
■ Disk partition information, including disk partition names, mapped logical volumes, mapped physical drives, and total capacity
■ Disk drive information: drive names, SCSI bus info, mapped disk partitions
■ Event information associated with each server
■ Real-time, historical, and forecasted capacity information for each logical volume
■ Real-time, historical, and forecasted performance information for various server and logical disk parameters
■ Ability to create and apply policies that notify and take action when new resources are discovered, or capacity utilization thresholds are exceeded
■ Numerous reports on asset and utilization information, exportable to HTML, XML, PDF, etc.