SAN appliances solve many of the problems that plagued first-generation implementations.
by ROBERT WOOLERY
Storage area network (SAN) appliances are accelerating the adoption of new business models with such benefits as lower total cost of ownership and increased Quality of Service (QoS). Moreover, they are delivering on the important benefit of competitive advantage with infrastructure flexibility and scalability.
SAN appliances are part of a technology trend seen in the network and Internet markets where technologies and products are initially developed to solve interoperability and scalability issues, then improve service and performance, and finally perform intelligent automated functions. This process follows the innovation curve of increased functionality with simplicity.
These new intelligent network infrastructure devices separate storage from servers and provide the means to cost-effectively design a low-latency, high-speed scalable storage network. They can provide simple manageability of heterogeneous storage pools in small workgroups and manage and transport vast quantities of rich media at very high speeds between multiple servers and clustered workstations and shared storage resources.
With the introduction of SAN appliances, IT organizations are starting to create custom-built infrastructures based on the specific needs of their customers and users, bringing the organization increased revenue productivity and accelerated return on investment (ROI). Other potential benefits of SAN appliances include ease of installation, centralized and virtualized management, multidimensional scalability, and increased QoS.
Why a SAN appliance?
SAN adoption began in earnest with the advent of Fibre Channel switched fabrics. However, systems administrators charged with installing and operating SANs quickly became disillusioned and skeptical about vendors' claims. SANs were not easy to implement and, in the early days, were fraught with interoperability problems. SANs didn't virtualize storage or enable the creation of virtual storage pools. Heterogeneous, cross-platform interoperability-particularly between Windows NT and Unix-was non-existent, and security was spotty and ineffective.
SAN appliances can provide scalable high-bandwidth, high-availability data access to shared storage resources, while supplying centralized data and network management in a single device. A SAN appliance enables a heterogeneous environment comprising standalone and clustered servers, workstations, and PCs with the ability to access and back up data stored in centralized or distributed storage devices.
A SAN appliance enables heterogeneous clients and servers to access data stored in heterogeneous storage subsystems and provides a centralized point of management and control.
SAN deployment has been driven by datacentric applications that take the form of "SAN islands" and by demanding requirements of rich media applications. As the deployment of SAN islands continues, companies will be faced with the task of interconnecting these islands to centralize storage management operations and create networked storage pools. These pools provide the critical functions needed to manage the growth in storage demand generated by streaming media, interactive e-business, content delivery, and high-performance computing. Through a SAN appliance's virtualization capability, information is intelligently managed across the network to transform physical storage devices on a SAN into networked storage pools that can be defined, resized, and managed without regard to operating system, server, or storage vendor.
A SAN appliance provides IT professionals with flexibility in scaling servers and storage on-the-fly, minimizing downtime and increasing productivity. It also supplies a level of security unavailable in existing SANs.
With a SAN appliance, data paths, switch configurations and topologies, storage devices, cache, and other hardware components become transparent to applications. The physical characteristics of storage and storage networking become transparent to the operating system through intelligence supplied by the appliance.
A SAN appliance typically resides between the hosts and storage pool, allowing IT professionals to virtualize all the storage and create logical unit numbers (LUNs) on-the-fly. These intelligent network infrastructure devices can also reside outside the data path to provide management and control of data and the network through a single appliance interface.
SAN appliances enable
- Storage sharing among multiple servers and clients;
- Host port and storage pool virtualization;
- Data and network intelligence;
- Scaling across multiple dimensions, with no reduction in data availability or performance;
- High-performance and network scalability; and
- Support for heterogeneous environment.
Two types of SAN appliances are the architectural approaches known as symmetrical and asymmetrical. A symmetrical SAN appliance sits in the data path, with data passing through the device. An asymmetrical appliance sits outside the data path. The debate lies in which architectural approach is better. However, this debate misses the real issue: solving the problem of data access.
Based on the application, data, or business model, either one or both SAN appliances may be the better approach. In some cases, both types of appliances working in concert create the most efficient data-access infrastructure by providing low latency, high-speed data access while easily scaling and maintaining a high level of QoS.
For example, you may want to use a combination of both symmetrical and asymmetrical SAN appliances in Internet deployments, rich media applications, or high-performance computing environments. The symmetrical approach provides the means to control and pump data throughout a switched fabric network. The asymmetrical appliance can manage and direct the flow of data, acting as a "control center" for the network. If you combine both architectures, you have a data access infrastructure that can move data quickly, while load balancing to meet changing demands.
Choosing a SAN appliance
Like many infrastructure decisions, choosing the right SAN appliance should start with an examination of your goals, which are affected by the application(s) and type and amount of data to be accessed, transported, stored, and managed. For example, if you're delivering rich media to users across the Internet, where QoS is important, a SAN appliance specifically designed for rich media and Internet applications makes sense.
If you're supporting transaction processing, you may want an appliance that's optimized for this data type and application. If you need a low-cost method to effectively manage a heterogeneous mix of storage systems and servers, then another type of appliance may be a better fit. Bottom line: Because various applications and data types have a wide range of different requirements and demands, a "one-size-fits-all" or general-purpose approach is less than optimal, reducing the potential benefits garnered from a SAN appliance approach.
Once data types and applications are determined, existing infrastructure and future performance and capacity needs come into play. For example, do you have file servers with storage you would like to incorporate into your network storage model, or do you need to support a series of high-performance workstations or servers, or both? Do you need to enable local-, metropolitan-, or wide area network connectivity? Are you planning to grow the number of users accessing data driven by the increase in content? Do you need to virtualize or share the data? As users or data increase, will I/O grow linearly or exponentially?
Once you have answered these questions and determined your needs, it's time to investigate vendors. The following is a checklist of things you may want to address in your selection process:
- For what data type and applications is the SAN appliance optimized?
- What dimensions can the appliance scale along?
- What are the QoS, data availability, and data-protection capabilities?
- What is the performance profile of the appliance?
- What are the data management capabilities (e.g., virtualization or data sharing)?
- What data and network security capabilities are enabled?
- What business continuance functionality does the appliance provide?
- Can it interoperate with other SAN appliances?
- Does it interoperate with other SAN connectivity devices?
- What host platforms does the appliance support?
- Does the appliance support legacy technology?
- Does it support other network storage models?
- Does it support other network environments?
- What storage devices does it support?
- What third-party software and hardware support is provided?
- What is the total cost of ownership?
- What level of infrastructure flexibility does the appliance provide?
- What competitive advantages can be created with the appliance?
The evolution of computer networking and the Internet have shown that computing markets gravitate toward highly intelligent devices that address and solve data-access problems simply. Intelligent SAN appliances are poised to bring significant total cost of ownership and ROI benefits and help create competitive advantage to all businesses requiring scalable data-access capabilities. The need for appliances that create intelligent storage networks from existing resources, coupled with the prospect of simplifying the upgrade to SANs, will accelerate adoption of new business models.
Robert Woolery is the vice president of corporate development and strategic planning at DataDirect Networks (www.datadirectnet.com) in Chatsworth, CA. He can be contacted at firstname.lastname@example.org.
Typical comments from systems administrators
If you have already installed a first-generation SAN or network-attached storage (NAS), the following comments from systems administrators may sound familiar:
"My NAS boxes supply only a fraction of the streaming performance I need today, and the capacity I'll need tomorrow."
"I needed a platoon of computer science PhDs to design and get my SAN running. Nothing works together unless I buy it from a single vendor."
"I want to mix and match the products from multiple vendors to get 'best of breed' solutions and leverage commodity pricing."
"Adding another server every time I need more storage is not my idea of scalability."
"My backup window is already too short. Backing up multiple NAS systems over my network just complicates the problem."
"Replicating data to solve my performance problem is very costly and a real management headache."
"Hiring and keeping system administrators is a real problem. If I could pool all my storage, I could manage it more effectively with fewer people."
The full potential of network storage will only be realized when these problems can be solved. Answers will come from new storage management software, virtualization, storage resource management, and device interconnect concepts. These concepts must be driven through intelligent infrastructure devices.
One example of the value a SAN appliance delivers is in Internet streaming media or content-delivery applications. The user experience of viewing media on the Internet is made up of several discrete processes that combine to deliver content to users. This process is called data access, and as in any process, the system is only as fast or responsive as its slowest element. For example, if you want to access data from a desktop in California, it may take five router hops to get outside your firewall. Once outside the firewall, it may take another 14 router hops to get from Los Angeles to Washington, DC, and then another five router hops from Washington, DC, to the site in New York. As you go through each router, you incur a latency toll as high as 86ms. In this example, the total latency toll was 1,634ms.
Once you're inside the New York Website, you need to transverse another five devices before you can access the data. From there, the storage network can add another 1,000ms to 3,000ms to bring the data to a streaming server and send it out through the firewall and back to you. The return journey for this particular data access request was 4,064ms. While you can record fewer hops and lower latency, this performance is not atypical. The average Internet request requires 15 to 20 hops and two to three networks (one way). Furthermore, the storage network segment of this process can impact data access performance by more than 50%.
The SAN appliance can be beneficial at the center of this data-distribution infrastructure. The SAN is at the heart of this process, moving data from its storage repository to the distribution grid. The SAN appliance operates as the power plant delivering and pumping data into this grid.
Once the data enters the grid via streaming servers or e-commerce processing units, the Internet network moves data to the user via a "bucket brigade" approach. This process may include moving from a large "core" data repository to small "edge" ones spread out throughout the Internet, repeating the process.
A rich media SAN appliance at the heart of a content-delivery system would reside in the core site and edge sites. Once content arrives at the core, the SAN appliance would store and forward the data to the edge sites where demand for the content exists.
The SAN appliance in the edge site would store the data and forward it to the end user. With its embedded intelligence, this device can reliably and efficiently stream data to users. It can deliver content at three times the performance and one-third the cost of existing network storage systems. Furthermore, most appliances have a set of APIs that lets providers customize their software and services to deliver a more robust set of value-added services.