Storage Area Networking (SAN)
The Solution for Network Backup
The continuing growth of managed data creates frustrating and often costly storage management problems for CIOs and network administrators. If your company is like most, the amount of data you have to manage nearly
doubles on an annual basis. Perhaps the most difficult problem to solve is determining how to back up the expanding volume of data.
The SAN Solution
SANs promise to alleviate the headaches associated with network storage backup by providing a 100-megabyte/second, high-performance data channel and shared, centrally managed storage devices. SANs will eliminate network
bandwidth concerns, substantially reduce administrative overhead, and reduce the total cost of ownership of the backup subsystem.
Driven by the emergence of gigabit-speed Fibre Channel networking, SANs offer
significant benefits to IT administrators. These benefits include:
- Data access flexibility
- Centralized data management
- Increased I/O performance
- Scalable network architecture
Problems with Conventional Backup
The high performance data path and the ability to share backup resources are obvious and immediate benefits of SAN solutions. Another key benefit enabled by SANs is flexibility in configuring the system. SANs
give you the ability to centralize backup operations for ease of management and greater utilization of the backup resource. In installations where centralized backup is not practical, SAN solutions provide the advantage
of centrally managing distributed backup resources. This flexibility gives you the opportunity to migrate your system from its current implementation to a more manageable configuration, and to scale the system as your
needs change over time.
The diagram (not shown) aboveillustrates how a SAN sub-network allows a tape library to be shared between multiple servers for LAN-less backup.
The network down time we used to call the backup window has all but closed due to the increased amount of data under management, limited network
bandwidth and the trend toward 24 x 7 operations. Traditional methods of managing backup, either in centralized repositories or in distributed server-attached backup, are no longer viable solutions.
Conventional backup solution: Centralized
Centralized backup limits the management overhead to a single storage repository. The problem isn't managing a centralized
backup repository; rather the challenge is getting the data to it. Conventional centralized backup solutions rely on the IP network as the data path. The problem with this solution is that the sheer volume of data
saturates network bandwidth. This results in long backup cycles that exceed the scheduled backup window. In fact, centralized backups often overflow into user uptime resulting in poor network response and generally
The diagram (not shown) above illustrates a typical centralized backup scenario where the LAN is the transport mechanism for the data to be backed up. This is not a viable solution due
to the combination of poor LAN data transfer performance and high data growth rates.
Conventional backup solution: Distributed
overcome centralized backup problems, most companies adopt a distributed backup configuration by attaching the storage directly to the servers. Distributed, server-attached backup doesn't suffer from network bandwidth
limitations but it does create administrative challenges since each server becomes a discrete storage repository or data island. It has become common for every server in an enterprise to have a tape drive installed and
for each server to be discretely backed up by that tape drive. This solution rapidly becomes impractical due to the need to touch every server every day and the fact that the servers rapidly outgrow the capacity of
installed tape drives.
The diagram (not shown) above illustrates a typical distributed backup solution. This type of solution is usually implemented to overcome network bandwidth limitations. This is not a viable
solution due to the lack of centralized management and high administrative costs associated with the management of multiple, discrete, backup operations.
Conventional backup solution: Combination
A third solution is to configure a type of hybrid solution where the network is subdivided into multiple centralized backup operations. This
type of installation eventually suffers from the problems inherent in both centralized and distributed operations.
The diagram (not shown) above illustrates a configuration combining aspects of centralized and
distributed backup operations. This type of configuration is common in larger systems and is often adopted as a compromise between the inability of the network to serve a centralized system and the high administrative
costs associated with managing backup on a server by server basis. Unfortunately these configurations tend to suffer from the negative effects of both centralized and distributed schemes.
All of the difficulties
imposed by conventional network backup approaches can be overcome with SAN solutions.
What is a SAN?
A SAN is a high-speed storage
network--similar to a LAN--that establishes a direct connection between storage elements and servers or clients. The SAN is an extended, storage bus which can be interconnected using similar interconnect technologies as
LANs or WANs: routers, hubs, switches, and gateways. A SAN can be local or remote, shared or dedicated, and uniquely includes externalized and central storage and SAN interconnect components. SAN interfaces are
generally ESCON, SCSI, SSA, HIPPI, or Fibre Channel, rather than Ethernet.
SANs provide a method of attaching storage that is revolutionizing the network because of the improvements in availability and
performance. In essence, a SAN is nothing more than another network, like a subnet, but constructed from storage interfaces.
SANs enable storage resources to be externalized from the server and in doing so, allow
them to be shared among multiple 'host' servers without impacting system performance or the primary network. Often referred to as the "network behind the server," SAN represents a new model that has evolved
with the advent of shared, multi-host connected enterprise storage.
This important technology is moving into the mainstream in distributed networking and will be the normal, adopted way of attaching and sharing
storage in the near future.
In addition to the fundamental connectivity benefits of SAN, the new capabilities facilitated by its networking approach enhance its value as a long-term infrastructure. These
capabilities further elevate the SAN's ability to address the growing challenges of data-intensive, mission-critical applications. The SAN environment complements the ongoing advancements in LAN and WAN technologies by
extending the benefits of improved performance and capabilities all the way from the client and backbone through to servers and storage. Some of the benefits of SANs include:
- High Bandwidth
- Modular Scalability
- High Availability Fault Tolerance
- Ease of Integration
- Reduced Total Cost of Ownership Fibre Channel
Although there are several possible SAN interfaces, Fibre Channel is the dominant technology implemented today. Fibre Channel is a one-gigabit per second data transfer technology that maps several common transport
protocols including IP and SCSI, allowing it to merge high-speed I/O and networking functionality in a single connectivity technology. Fibre Channel is an open standard as defined by ANSI and OSI standards and operates
over copper and fiber optic cabling at distances of up to 10 kilometers. It is unique in its support of multiple inter-operable topologies including point-to-point, arbitrated-loop and switching, and it offers several
qualities of service for network optimization. With its large packet sizes, Fibre Channel is ideal for backup, video, and graphic or mass data transfer applications.
The SAN Environment
Historically in storage environments, physical interfaces to storage consisted of parallel SCSI channels supporting a small number of SCSI devices. With Fibre
Channel, the technology provides a means to implement robust SANs using switched and arbitrated loop topologies that consist of hundreds of devices. Fibre Channel SANs yield a capability that supports high bandwidth
storage traffic on the order of 100 MB/s. Enhancements to the Fibre Channel standard will support even higher bandwidth in the near future.
Depending on the SAN implementation, several different components can be
utilized to build the physical SAN. In smaller SAN environments, Fibre Channel arbitrated loop topologies employ hub and bridge products. As SANs increase in size and complexity and are constructed to address higher
availability, switches or fabrics are introduced. Storage subsystems and servers are attached to fabrics or loops using Fibre Channel adapters that must also provide a management capability.
Due to the varying
scale of SAN implementations outlined above, it is useful to view a SAN from both a physical and logical standpoint. The physical view allows the components of a SAN to be identified and the associated physical topology
between them to be understood. Similarly, the logical view allows the relationships and associations between SAN entities to be identified and understood.
SAN Physical View
From a physical standpoint, a SAN environment typically consists of four major classes of components:
- End-user platforms such as desktops and/or thin clients
- Storage devices and storage subsystems
- Interconnect devices such as switches, hubs, bridges, and adapters.
SAN Logical View
Normally, network facilities based on traditional LAN and WAN technologies provide connectivity between end-user platforms and server components. However in SANs, end-user platforms may be attached to the fabric
and may access storage devices directly.
Storage subsystems are connected to servers, to end-user platforms, and to each other using the facilities of a Fibre Channel fabric or backbone. Fabrics are composed of
one or more fabric elements that provide transport services by routing frames between connected entities. Fabrics may further be broken down into logically independent sections called sub-fabrics for each possible
combination of data rate and class of service. Sub-fabrics are again divided into regions and extended-regions based on compatible service parameters. Regions and extended regions can also be divided into partitions
called zones for administrative purposes.
The diagram (not shown) above illustrates one of many possible storage sharing scenarios enabled by SAN technology. In this example all of the servers can be backed up
over the SAN and the direct SAN attached storage can also be managed as part of a centralized backup to the shared tape library.
From a logical perspective, the SAN is defined by software rather than hardware topology. The logical operation of a SAN requires applications and management tools that are aware of--and capable of sharing--storage
resources among many host systems. The logical management architecture includes several layers from data management applications down to device management, and must include administrative control at each layer.
The leading tape software vendors have addressed sharing automated tape libraries in SAN environments by implementing layered architectures like the one depicted above. The device management layer virtualizes the backup
resources so they become dynamic from the applications' point of view. This allows SAN-aware applications to share backup resources among multiple processes. It also allows the resources to be local or remote and
centralized or distributed while still being managed centrally. This implies nearly unlimited physical connection permutations. An advantage of this is that the storage system can easily be scaled as storage needs grow.
Exabyte has worked closely with these vendors to certify their software solutions in a variety of configurations. We have also addressed the need to independently and remotely manage the tape libraries and their
respective Fibre Channel controllers by developing browser-based monitor applications for Exabyte Fibre Channel routers and tape libraries.
Exabyte's SAN management software allows you to configure devices and
monitor operations in the SAN environment through a user friendly browser interface.
Solving the Backup Dilemma
promises to be the storage management architecture for the future; however, it is still a young solution. There are numerous committees working to establish standards for SAN management in the same way LAN/WAN
management has been standardized over the past two decades.
Many SAN pundits will present a complete SAN architecture including shared file systems, shared disk subsystems and, of course, shared backup systems.
While this is an accurate depiction of what will be possible in the future, most of these discussions don't address how to evolve an existing system to a SAN-based storage solution.
Employing a SAN to resolve
backup difficulties will help achieve many goals including: eliminating backup window constraints, centralizing backup operations and management, effectively managing distributed systems, and automating many operations
which currently require substantial human intervention.
Storage networking, including SANs, represents the future of data
centric computing. Although some aspects of storage networking are still in the development phase, Exabyte has developed, tested, and proven solutions for the most common network backup concerns.
If you are
planning to implement storage networking and you are already struggling to solve your network backup problems, you should begin by building a SAN to address those problems. Your backup SAN will help you overcome
conventional network backup problems and provide the foundation for your storage area networking needs in the future.
© 1999 Exabyte Corporation. All rights reserved.
EXABYTE is a registered trademark of Exabyte corporation. All other trademarks are the property of their respective companies.