Storage area networks (SANs) offer advantages to companies seeking to meet the constantly changing data storage requirements of today’s large, rapidly growing networks. SANs provide high-speed, scalable, networked storage solutions that separate the storage devices and storage access from the enterprise network. This provides the opportunity to centralize storage management and to relieve storage-related bottlenecks in the LAN and within servers. SANs can solve the problem of any-to-any connectivity between storage devices and servers. A SAN also offers new approaches to storage, such as disk and tape pooling, heterogeneous data sharing, and off-the-network server-less backup and restore.
While a SAN can significantly improve data and storage device management, its actual performance depends on a number of factors. The first is the quality, features, and interoperability of the storage connections and devices (e.g. Fibre Channel host bus adapters, hubs, switches, bridges, routers, and cables) that are employed in the SAN. With inferior components, it can be difficult to maintain data availability. With such devices, it is often necessary to spend an inordinate amount of time trouble-shooting problems, re-routing data paths, and installing components. It is also unlikely that one manufacturer will be able to provide the best features and quality for every SAN hardware and software component. This makes it highly desirable to choose the best available components for each function within the SAN.
Just as important as the quality and feature set of the devices is the effectiveness of the SAN management application in tying these devices together and simplifying the complexity of the storage network. IT managers need to configure, monitor, and maintain the SAN, as well as plan for, and accommodate, growth. Without a robust management application designed to meet these requirements, even the best SAN installation can be an administrative nightmare.
To ensure that each component in a SAN provides the desired functionality and quality, it makes sense to select “best-of-breed” components. What represents the best product for one installation may not always be the best choice for another. Requirements may vary significantly, particularly in the areas of availability, capacity, performance, and cost. A SAN that consists of “best-of-breed” switches, hubs, routers, storage subsystems, cables, connectors, and software is therefore most likely a heterogeneous, multi- vendor network.
Because it is common to add components to SANs as specific needs arise, it is also typical to have a heterogeneous network with components of varying quality and capability. For economic reasons, it may not be desirable to replace older components in a SAN until they fail or become obsolete, but it is still necessary to manage the network.
A heterogeneous SAN based on “best-of-breed” devices and a mixture of older and newer components can present significant interoperability and management problems because the storage industry lacks common standards for devices and management software. While some standards, such as the FibreAlliance management information base (MIB) are emerging, these standards are not fully implemented by many of the leading hardware vendors. Instead, vendors often employ their own, proprietary MIBs and application programming interfaces (APIs).
Addressing the challenge
How can you avoid the interoperability and management problems that can occur in a heterogeneous (multi-vendor) SAN? The best way is to make device interoperability a consideration during the component purchase cycle, and to select a SAN management application that provides flexibility in managing with current and forthcoming standards, features, and the proprietary interfaces of all the major equipment and software vendors.
Standardizing on one best-of-breed ven-dor for each type of SAN component can help to minimize the complexity of heterogeneous SANs. For example, standardizing on one switch vendor for each SAN installation provides the ability to cross-connect the devices for improved availability in the event that a switch fails. However, even if the SAN installation includes only one vendor per component type, interoperability between the various types of components in the SAN must be addressed. Most leading vendors have interoperability certification programs, through which product interoperability is tested extensively. Ensuring that the selected SAN components interoperate effectively in these labs through the applicable certification programs will reduce the likelihood of device interoperability problems.
Once the SAN devices are chosen, selection of the right SAN management tool is key. This application should be flexible, portable, and scalable. In addition, it should be simple to use and protect the SAN from novice administration errors.
Flexible SAN management applications offer some type of plug-in architecture, or modularity, so that new devices can be quickly incorporated into the software. The management software should also be able to easily support new discovery protocols. And the software should have a layered architecture so that it is possible to easily integrate any feature in the SAN management application with the APIs of other vendors’ products.
A layered architecture can also reduce the amount of code that any modification (e.g., feature addition) impacts, because functions are contained within specific modules rather than distributed throughout the application. This simplifies software maintenance and reduces the chance of a code modification introducing new problems in the application. Monolithic software architectures that do not provide this flexibility increase the chance of debilitating application bugs and interoperability issues.
Each of the hubs, switches, tape libraries and other devices in a SAN may have its own management software. Ideally, to keep from having to monitor hundreds of consoles, these devices should be managed from a single console. Most SAN management applications allow for on-site or remote SAN management.
To simplify SAN management in a heterogeneous environment, management applications typically make it easy to operate software tools from a variety of vendors. Automatically launching native software tools for any of the devices (or management functions) in the SAN makes possible unique device vendor error- reporting and correction tools and enables the software to scale with added devices and features.
When a manager needs to add a device from a new vendor, the SAN management application enables association of the new vendor’s devices with the desired software. A simple launching mechanism enables the tool to find and launch the unique vendor application.
Storage management applications can also provide integration with enterprise-level network management applications, in case these applications are someday added or merged with the other administration tools. This should include, at a minimum, pass-through of SNMP traps (events). Some management applications can also forward events to multiple “listeners” (e.g., for remote location monitoring), provide topology import/export, and bill of materials export in standard formats.
Management software for heterogeneous SANs is usually portable to a variety of operating systems. Enterprise networks often have Windows NT and one or more versions of Unix. SANs often originate with the Unix-based storage requirements of development groups, and later expand to include NT-based administrative storage.
SAN management software has to scale with storage requirements, which in a typical enterprise are growing at more than 50% a year. Many management tools allow for dynamic discovery so that SAN components can be added and deleted without having to re-start the SAN. Most vendors also supply a mini-map or alternative method (e.g., zoom levels) for navigating through large topologies.
The application should also provide a hierarchical user interface to ensure scalability as the SAN grows. If the abstraction levels are not built into the software, problems with scalability are probable later on.
At a minimum, SAN management applications may support out-of-band discovery. If the SAN experiences a catastrophic failure, remote out-of-band discovery and monitoring can ensure that the management application is still able to notify the administrator of the failure.
The appearance and ease-of-use of the user interface can significantly affect the way a storage network is managed. An interface that is intuitive and easy to use can greatly simplify advanced management functions for administrators and provide a link between the storage network and other enterprise management tools. For example, each switch and router vendor has its own method of constructing server storage zones. The user interface can simplify zoning by providing an intuitive, visual method of constructing zones, while relegating the complexity of the different zoning methodologies to the lower (and user-transparent) levels of software.
The network topology should be displayed on the screen in a way that makes sense to network/storage administrators. For example, it helps if the LAN devices are displayed in a different color and grouping than the SAN devices. For easy identification, errors can also be displayed in color and categorized based on their severity.
Managing a heterogeneous SAN
Given a SAN management application that fulfills these requirements, what should the application do for users? To tie the diverse SAN devices together in a meaningful way, SAN management applications may provide:
- Automatic device discovery
- Topology mapping
- Status monitoring
- Event logging and notification
In addition, SAN management software sometimes allows administrators to manage advanced functions, such as backup, zoning, and storage pooling (virtualization). The storage management application can provide these services either directly or through APIs to other applications/devices.
Organizations often need to support multiple subnets. SAN management applications are usually able to automatically discover devices on any subnet, through local or remote monitoring. This type of application can also display multiple subnets on a single management console. Given the subnet address, the software should be able to support the added subnet immediately, without complex configuration tasks.
After obtaining the subnet address, the management software automatically discovers all the devices on the SAN-hubs, switches, HBAs, routers, etc.-and their connections. The application identifies the device vendors, worldwide names, and other properties if they are available from the device software. If the device is not an identified component, management applications often provide the capability for a user-defined device, making it possible to create a complete SAN component map. The characteristics of each component are usually available by clicking on the component icon, and can be saved so that the user only enters information once.
SAN management software also maps device locations in the network topology. The topology information can be presented in the form of a visual map that includes all the components, devices, and their interconnections, showing groupings of the hosts (or HBAs) and switches.
To provide users with an intuitive understanding of the network, redundancies should be displayed graphically. In the past, this type of visualization, or diagramming of the SAN topology, was an arduous task that needed to be carried out manually.
When a SAN device is added, moved, or removed, the software must be able to detect it, and adjust the topology display accordingly. Devices and connections that are removed or failed are typically displayed differently (e.g., in red) to allow trouble spots to be identified quickly.
The discovery and mapping process becomes more complicated when a device uses a vendor’s private discovery protocol. The software must still be able to detect these devices, and add them to the topology.
As the SAN management application discovers and maps the devices, it should save the information so that it is “aware” of any changes that occur in the SAN. This feature is called persistence.
All user-entered information is also saved. Persistence, combined with im-port/export capability, offers the user flexibility in using the software to identify changes in the SAN (e.g., between administrator shifts), and to re-create saved configurations (e.g., if a SAN needs to be quickly re-established after a company move or local catastrophe).
Management applications often monitor the SAN for changes in device and connection health. Out-of-band discovery via SNMP features the ability to “listen” for device traps (events), such as a power supply failure or environmental problem. If a catastrophic event occurs, however, such as a device failure or a cut/broken cable, the component may not be able to send a call for help.
The management application, therefore, periodically polls the devices to make sure everything is still functional. Ideally, polling intervals and timeout periods are changeable by the user. Together, event listening and polling can determine the overall health of the entire SAN.
Monitoring services are becoming increasingly important for SANs. A growing number of organizations (such as storage service providers) monitor other organizations’ storage network or co-locate storage around the world. SAN management applications provide remote monitoring capability, including event forwarding services and remote notification of failures. All events should also be logged for a record of a SAN’s health and performance. Some applications enable selected event types to be aggregated at user-specified time periods, and sent via SMTP to pagers or e-mail locations.
Backup, zoning, pooling
Advanced function support may also be provided in the management software itself or through the ability to launch other software modules. Management software can be used to launch functions from a particular device so that users can create virtual appliances. For example, a third-party backup application might be associated with a particular host, or even a NAS box on the front end of the SAN, and the launching host could be renamed “backup appliance.”
One major advantage of a SAN is its ability to organize storage devices and data paths in many different ways to effectively manage the transport and storage of data. One important configuration issue to address is data-path zoning. This is primarily a concern if more than one operating system exists in the SAN.
To prevent one rogue operating system from “taking over” all the available storage, the storage can be divided, or zoned, for each operating system. Additionally, administrators may need to segregate sensitive data so that it can be only accessed by a particular host. Because vendors’ approaches differ greatly, it may be advantageous if the management software provides this service.
SAN management software that supports data-path zoning should provide an intuitive interface that allows managers to select both the host and storage ends of the data path. For example, the software might enable managers to highlight the particular path of interest between two devices, and enable or disable it.
An alternative to zoning for multiple operating systems is to pool the storage. By using universal file system software or specialized SAN controllers, the administrator can enable multiple operating systems to share data. The management application should have the flexibility to allow users to implement storage pooling and zoning methodologies through a simple user interface.
The management of heterogeneous SANs requires due diligence in component selection, including both hardware and software. Selecting best-of-breed hardware and software that can handle such a complex environment will simplify SAN management, while providing enterprise-wide data protection.