Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Storage switch growth spurred by new features

A switch separates a network into segments. This helps to manage network traffic by passing traffic at wire speeds to only the segment(s) that host destination nodes. But, network needs vary dramatically, and switches have proliferated in size and features -- often adding intelligence that can support high-level capabilities like storage virtualization, security, and a variety of other tasks.

Vendors and product selection [continued from page 1]

QLogic's SANbox 3050 is another entry-level SAN switch offering eight 2 Gbps FC ports and management software featuring a wizard-driven GUI. The SANbox 5600 is a stackable switch from QLogic, starting with eight 4 Gbps FC ports and expanding to 16 ports in four-port increments. Four high-speed ISL ports allow fast inter-switch connectivity -- four switches can be stacked, allowing up to 64 ports in total. Like the 3050, the stackable 5600 is also configured and zoned with simple wizard-driven management software.

In the mid-range, blade switches are readily available, but are typically intended for specific blade server chassis models. For example, McData's 4314 and 4416 FC switch models support the Dell PowerEdge 1855, sporting a total of 16 FC ports running up to 4 Gbps. A McData FC blade switch module with 10 or 20 4 Gbps FC ports can be purchased for the IBM eServer BladeCenter. Optional security software is available to help manage data access. The HP p-Class BladeSystem is served by a 10 port, 4 Gbps McData SAN blade switch with features like ISL trunking, zoning, binding and authentication. As with edge products, the blade servers tout ease of installation, configuration, operation and maintenance.

Another mid-range switch product is exemplified by Brocade's SilkWorm 4900 with up to 64 ports offering 4 Gbps FC connectivity in a 2U enclosure. The 4900 can be expanded in 16-port increments and offers high-availability features like non-disruptive software upgrades and redundant hot-swappable components. Management can be accomplished across the network using any Web browser with SSL. The large port count makes the 4900 ideal for switch consolidation and fabric simplification.

Numerous director-class switches are available at the high-end of the switch spectrum. Cisco's MDS 9513 is one of the largest director switches available, sporting as many as 528 fibre channel ports in a single chassis -- each port capable of 1, 2, 4, or 10 Gbps. The 9513 includes support for VSANs and VSAN routing, along with advanced security features. Brocade's SilkWorm 48000 SAN director scales non-disruptively from 32 to 256 active 4 Gbps FC ports in a single domain, supporting enterprise management, high-volume transaction processing, and backup tasks. Features like continuous monitoring, non-disruptive upgrades, and redundant hot-swappable components ensure high availability. Centralized management supports heterogeneous device connectivity within a variety of server and storage platforms and operating systems. McData's Intrepid series of directors also support enterprise-class SAN switching. The i10K director provides up to 256 FC ports running at 2, 4 or 10 Gbps, depending on the service modules that are installed. Redundant components and hot swappability allow for availability rated to 99.999%, while a variety of management options let administrators configure and maintain the switch using existing management tools.

Selecting the right product

Choosing the best switch for your environment can be challenging. Considerations like compatibility, expandability and intelligence are often the most important factors, but issues like cost and existing vendor relationships can frequently influence the choice. Once you narrow the field, a few potential candidates can be thoroughly tested in-house. Analysts suggest the following points that can help you identify the best product for your own production environment.

Evaluate the switch's support in your data center. Although switch interoperability issues have largely been resolved, analysts recommend checking the switch to verify that it supports the elements of your specific environment. "Support for the storage devices, and the operating systems, and the drivers that you have is still something that needs to be checked," Garrett says, noting that EMC had trouble interoperating with early McData i10K director-class switches. This delayed the broad adoption of McData switches in EMC environments for up to six months before EMC could adequately support them. "The fact that it wasn't on the interoperability matrix and supported by the storage vendor yet caused them to wait until it was."

Consider the available switch ports and bandwidth. A switch will need an adequate number of ports, and should offer adequate scalability -- either by licensing and activating additional (spare) ports or by interconnecting additional switches. Each switch port should also provide adequate bandwidth to support the anticipated network traffic. When interconnecting switches, analysts emphasize the importance of ISL (inter-switch link) traffic since moving data between switches can introduce unwanted latency to the network.

Leverage additional switch intelligence on the network. "Increasingly, users are looking to centralize the configuration, the monitoring and the services that are running right now on storage systems and servers and instead have them run on the network," Garrett says. Hot data migration (moving data while the application runs), centralized LUN management, virtualization, replication and copy services are just a few of the tasks that can be centralized onto the network and handled through advanced switch devices. Users considering network management consolidation should pay close attention the vendor's roadmap for future feature support.

Consider switch cost and commoditization. Switches are experiencing significant price pressure due to intense vendor competition, and new products are quickly becoming commodities. This provides appealing purchase opportunities for established switch products, but also helps make advanced features more affordable. "One of the things that's driving price also is the arrival of four gig [4 Gbps Fibre Channel]," Schulz says. "As the four gig volumes ramp up, we should see more price reduction -- or be able to buy more for the same amount."

Best practices for implementation

Switches are at the core of most network infrastructures, so it's particularly important to document any installations and carefully track any changes to the switch setup or configuration. Switch upgrades or reconfigurations are also excellent opportunities to address congestion or bottlenecks that may have evolved over time. Another common trap to avoid is over-consolidation -- foregoing critical network resiliency in favor of simplicity. Some well-grounded policies can help smooth any switch implementation issues.

Implement change control in the network. Undocumented changes can eventually make it impossible to troubleshoot or upgrade the switch and its configuration on the network. An organized process must be invoked to manage changes as they occur. "Have a change control process in place that is understood and documented so that you can plan for storage networking changes," Garrett says. "And have the change documented and auditable afterwards."

Take the time to plan and document switch deployments. Switches are the focal points in many networks, so proper implementation must include adequate pre-installation planning. It's important to define the physical interconnections that are required, but also the switch's logical setup and configuration. Proper preparation will always make the actual deployment proceed far more smoothly. Once the installation and configuration are completed, the new deployment should be completely documented for future reference -- documentation facilitates future troubleshooting, and forms the basis for change management.

Configure switches for availability and resilience. When considering the introduction of new switches to the data center, analysts underscore the need to plan for growth, plan for upgrades and plan for inevitable maintenance. An important part of that is to design deployments that eliminate single points of failure. "For example, if you were going to deploy a director to consolidate multiple smaller switches, don't go a single director; deploy a pair of them," Schulz says. "Avoid that temptation to over-consolidate."

Design switch deployments to remedy traffic congestion and bottlenecks. Bottlenecks are common in networks -- especially when networks experience significant growth in a short period of time. Analysts suggest that any switch deployment should be preceded by an evaluation of network bottlenecks. In many cases, points of congestion can be addressed and remedied during switch deployments or upgrades. "The key thing is [to] design for the applicable level of performance that you need both for normal and for peak processing workloads," Schulz says.

Implement security measures. As switches assume more control over network behaviors, security is becoming a greater concern. Secure any switch management tools by changing default passwords, and make the effort to change passwords regularly. Secure (disable) unused ports to prevent onsite hacking. Set policies and procedures that limit access to switch management tools. ***

[Return to previous page]

Dig Deeper on SAN technology and arrays

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.