Gigabit over copper promises an affordable network performance boost

These are auspicious times for companies looking for faster network performance. Because of competition among vendors that are steadily increasing their 1000Base-T switch offerings, Gigabit Ethernet is more affordable and easier to deploy than ever. Unfortunately, assessing which Gigabit Ethernet solution is best suited to your needs isn't always easy. Vendors advertise their switches with the customary performance claims and glossy marketing brochures, but neither gives you a definitive idea of how a certain switch will perform in your particular environment, much less how easy it is to set up and manage. To help you get some inkling of what to expect from the latest 1000Base-T switches, I brought some of them into the Windows & .NET Magazine Lab for a test drive.

At the time of writing, at least a dozen vendors offered a wide range of Gigabit Ethernet switches. To narrow the field, I looked only at managed switches that offer native 1000Base-T ports or can accommodate a Gigabit Interface Converter (GBIC) module for uplink ports. I also looked for switches that represent a cross-section of price and functionality. And because I wanted to simulate a typical network-upgrade scenario between a server room and a remote wiring closet, I looked for vendors that could deliver Fast Ethernet to the desktop with 1000Base-T uplinks that I could pair with a 1000Base-T backbone or aggregation switch. My search led me to the most recent Gigabit Ethernet switch offerings from 3Com, Cisco Systems, and Dell. For those looking for a chassis solution, I tested a Hewlett-Packard (HP) chassis switch with Gigabit and Fast Ethernet switch modules.

The testing was fairly simple. I first conducted a baseline test on the Lab network, in which I performed routine administrative tasks (e.g., software-application deployment) while monitoring latency-sensitive streaming-media applications. I next installed Gigabit Ethernet NICs in the network's servers and in some desktop machines and went through the usual regimen of performance tuning. To simulate an upgrade scenario, I then replaced older Fast Ethernet switches in the Lab's test network with each vendor's Gigabit Ethernet offerings. As I worked with each vendor's switches, I noted how easy they were to configure and manage from basic setup to advanced features such as Virtual LAN (VLAN) configuration, routing, and cost-of-service configuration. Finally, I performed the same tasks and ran the same software as I had in the baseline test to check for a performance improvement.

3Com SuperStack 3
3Com sent a pair of 3Com SuperStack 3 switches for testing. SuperStack 3 Switch 4900 is a Layer 3—capable (with firmware upgrade) aggregation switch. Layer 3—capable means that the switch can also act as a router, implementing switching algorithms based on IP address as well as media access control (MAC) address. The IP address is at the Network layer, or Layer 3, of the Open System Interconnection (OSI) reference model; the MAC address is at the Data Link layer, or Layer 2. Aggregation means that Switch 4900's 12 autonegotiating 1000Base-T ports work well as an aggregation point for other Gigabit Ethernet links, whether they be lower-tier switches or high-performance hosts. Switch 4900 has space for an expansion module that lets you add four additional 1000Base-SX or 1000Base-T ports. It also has a connector for a proprietary 3Com redundant power system.

SuperStack 3 Switch 4400 is 3Com's most recent offering for providing Fast Ethernet connectivity to the desktop along with manageability features. It has 24 autonegotiating 10/100Base-T ports and two slots for modules that provide additional media options or stacking capability with other 3Com switches. As many as eight Switch 4400s can be stacked and managed as one cluster. To provide a Gigabit Ethernet uplink to Switch 4900, I had to install a 1000Base-T module in one of the two available slots.

At about $416 per Gigabit Ethernet port, Switch 4900 is in the middle of the price range of the equipment I tested. The combination of the two SuperStack 3 switches would work well in a network hardware—upgrade scenario like the one I simulated. Switch 4400 could simply replace an older Ethernet switch or hub in a wiring closet, and its gigabit uplink port could connect, through existing Category 5 copper cable, to a Switch 4900 aggregation switch in a server room.

3Com SuperStack 3 Switch 4900
and 3Com SuperStack 3 Switch 4400
Contact: 3Com * 408-326-5000 or 800-638-3266
Web: http://www.3com.com
Price: Switch 4900:
$4995 for 12-port 100/1000Base-T copper managed switch;
Switch 4400: $1900 for 24-port 10/100Base-T
managed switch with Layer 3 functionality and 1000Base-T uplink module

Cisco Catalyst
Cisco Systems' Cisco Catalyst 3550-12T and Cisco Catalyst 2950T-24 switches are also a well-matched pair for the network-upgrade scenario. The Catalyst 3550-12T is a Layer 3 aggregation switch with 10 autonegotiating 10/100/1000Base-T ports and two hot-swappable GBIC slots that you can use for stacking or to provide an additional two Gigabit Ethernet ports. The Catalyst 3550-12T also has a connector for a Cisco redundant power supply. The cost per Gigabit port, nearly $1000, puts this switch at the high end of the price range but is, at least in part, a function of the switch's high level of built-in functionality. If you manage a large network, the extra features, such as fully dynamic IP routing and advanced Quality of Service (QoS) options, might be well worth the cost.

The Catalyst 2950T-24 is a 24-port autonegotiating 10/100Base-T switch with two autonegotiating 10/100/1000Base-T uplink ports. This switch, like its big brother, uses a modified version of Cisco's Internet Operating System (IOS) 12.1. Thus, any experience you might have with Cisco IOS is helpful in configuring these switches.

Cisco Catalyst 3550-12T and
Cisco Catalyst 2950T-24
Contact: Cisco Systems * 408-526-7208 or 800-553-6387
Web: http://www.cisco.com
Price: Catalyst 3550-12T:
$9995 for 10-port 10/100/1000Base-T copper
managed switch;
Catalyst 2950T-24: $2395 for 24-port
10/100Base-T managed switch with two Gigabit Ethernet copper uplinks

Dell's PowerConnect
At the time of writing, Dell had just thrown its hat into the Gigabit Ethernet switch ring with the release of its PowerConnect line of switches. These switches are the least expensive of those I tested and thus should be popular with small to midsized organizations that want to deploy Gigabit Ethernet but don't need Layer 3 functionality. Indeed, the Layer 2—capable PowerConnect 5012 provides 10 autonegotiating 10/100/1000Base-T ports for only $140 per port. The PowerConnect 5012 also has two GBIC slots for fiber uplinks.

The PowerConnect 3024 is Dell's managed Fast Ethernet desktop switch. Like the other vendors' desktop switches, the PowerConnect 3024 provides 24 autonegotiating 10/100Base-T ports. On the back, the switch has two built-in Gigabit Ethernet stacking ports that you can use with proprietary cables to stack an additional five switches. It also has two uplink ports that are 10/100/1000Base-T by default but can be used instead with fiber uplink modules that you purchase separately. Like the products from 3Com and Cisco, the Dell switches are well suited to upgrade an existing Fast Ethernet network.

PowerConnect 5012 and PowerConnect 3024
Contact: Dell * 800-234-9999
Web: http://www.dell.com
Price: PowerConnect 5012:
$1399 for 10-port 10/100/1000Base-T copper managed switch with two GBIC ports; PowerConnect 3024: $756 for 24-port 10/100Base-T
managed switch with two Gigabit Ethernet copper uplinks
(switchable to fiber with separately purchased modules)

HP Procurve
For certain environments, a chassis switch is an alternative to consider over stacking modules. Chassis switches offer flexibility because you can add or replace individual modules, or blades, to meet changing requirements over time without replacing the entire switch.

HP supplied its HP Procurve Switch 4108GL mini chassis for testing. This switch provides eight slots for modules that you can mix and match as you need. HP offers standard 24-port 10/100Base-TX modules as well as Gigabit Ethernet modules for copper and fiber. As is typical with chassis switches, the initial per-port cost is high (more than $1000 if you buy the chassis and just one 6-port Gigabit Ethernet module) but goes down as you approach the switch's full capacity.

Switch 4108GL uses HP's Fast Path Technology, which means that each of its modules is an independent switch. Thus, you can upgrade modules without replacing the entire chassis. For example, Switch 4108GL is a Layer 2 switch, but you can replace a module to provide the chassis with Layer 3 functionality. The independent modules can also forward packets between their ports without crossing the central switching fabric of the chassis—an ability that cuts down on traffic.

Switch 4108GL was easy to set up and configure and offers similar management capabilities to the other switches I tested, including full SNMP support and a Web interface. Also included with a switch purchase is HP's Toptools for hubs and switches.

HP Procurve Switch 4108GL
Contact: Hewlett-Packard * 970-635-1000 or 800-752-0900
Web: http://www.hp.com
Price: $4239 for chassis switch with eight slots;
$1869 for 6-port Gigabit Ethernet copper module;
$1699 for 24-port Fast Ethernet module

Installation, Setup, and Configuration
The physical setup of the switches in the Lab test network was as expected. All the desktop switches are 1U (1.75"), as is the Dell aggregation switch. The 3Com and Cisco aggregation switches are 1.5U (2.625"). All the switches use typical rack-mount hardware. The Dell and 3Com switches offer a nice feature called AutoMDI/MDIX, which lets them automatically sense and connect to a cable whether it's wired straight-through or crossover. This feature eliminates the common scenario of an administrator manually assembling a crossover cable and introducing potential faults.

My experiences installing and configuring the Gigabit Ethernet switches were varied. Several aspects of initial setup were common, such as using a PC and terminal program to connect to a terminal port on each switch to configure basic settings for management and remote access. After that point, however, the Cisco switch was definitely more difficult to configure than the Dell, 3Com, or HP switch.

The Dell, 3Com, and HP Gigabit Ethernet switches are easy to install: When you plug them in and connect your network devices to them, they immediately go to work as basic Layer 2 switches. To configure advanced options on these switches, you can connect directly to the switch through the console port or remotely through a Telnet session or Web browser. (The 3Com switch also can automatically get an IP address from a BOOTP server instead of requiring you to use the switch's console port to assign the address.)

I found that I could follow the menus and use a little bit of trial and error to configure most of these three switches' options. Some of the advanced Layer 3 options on the 3Com switch forced me to use the manual a few times, but the vendors did a good job making their text- and Web-based interfaces intuitive and easy to use. The only glitch I ran into was a mysterious memory leak that caused poor performance on the host PC I used to manage the Dell switch. Dell, like the other vendors, uses JavaScript in its Web interface, but it's the only vendor that doesn't require you to install Sun Microsystems' Java Virtual Machine (JVM). Thus, my problems might have been related to Microsoft's JVM.

An added bonus with the 3Com switch is its Network Supervisor management software. This application, which 3Com provides at no additional charge, offers a range of features that might appeal to smaller organizations that don't have a network-management application. Network Supervisor can perform network discovery, monitoring, and alerting and reporting functions, as well as let you centrally manage and upgrade 3Com switches.

HP's switch was easy to set up and offers hot-swappable modules. As is characteristic of a chassis solution, Switch 4108GL has a great deal of flexibility to accommodate the rapidly changing needs of many businesses. You could, for example, choose to populate the majority of slots with 6-port 1000Base-T modules for an aggregation switch. HP also offers a stacking module that lets you daisy-chain its switches if you need more capacity.

As I mentioned, setting up the Cisco Gigabit Ethernet switch was a bit more complicated. When I tried to just plug in the switch and plug in my hosts, the switch didn't respond. Cisco includes documentation on the CD-ROM that comes with the switch, but Cisco's document reader wouldn't install, so I couldn't read the nonstandard document formats. Thus, I was forced to download the manuals in PDF format from the Cisco Web site. I've always appreciated Cisco's excellent Web-based support and thorough documentation, but I had to invest a few hours in the 700-page manual and the IOS command line before I had the switch configured and running correctly. Ultimately, setting up the switch was similar to setting up a Cisco router, which makes perfect sense when you consider that that's what a Layer 3 switch really is. Yet, I felt that Cisco could have made the initial setup easier.

After I had the switch running, I used Cisco's Web-based Cluster Management Suite (CMS) for further configuration. CMS lets you manage clusters of as many as 16 Cisco switches as one IP address. The Web interface was fairly intuitive and was well suited for several tasks, such as setting up VLANs and QoS prioritization. That being said, the IOS command line is still the best interface for configuring the many detailed options that make the Catalyst 3550-12T both powerful and flexible. Besides, most of the documentation is dedicated to performing tasks from the command line, whereas only one chapter of the manual describes using CMS.

The four vendors' Gigabit Ethernet switches have similar management capabilities. All of them provide firmware upgrades through Trivial FTP (TFTP)—something I had to test immediately on the 3Com switch to upgrade it for Layer 3 capability. All the switches provide SNMP agents and support similar sets of MIB extensions for SNMP-based management applications and Remote Monitoring (RMON). In addition, the switches provide the ability to monitor switch activity through Web clients or even with bundled network-management software, such as 3Com's Network Supervisor. Unless you're running a small network, however, you'll likely be using some form of network-management application that supports SNMP; all four switches are well suited to SNMP-based management and monitoring.

Gigabit Performance
My goal in testing Gigabit Ethernet switches in the Lab was to get an idea of real-world performance gains rather than perform benchmark testing that would pit one product against another. To simulate a production network, I set up the Lab test network as follows.

The network consisted of 20 Fast Ethernet client PCs running Windows 2000 Professional and Windows NT Workstation and six Win2K servers. I first established a baseline by using these machines on the existing Fast Ethernet switches in the Lab. I recorded average throughput and elapsed time to perform certain tasks and monitored general network responsiveness during peak loads. After I established some numbers for comparison, I installed 64-bit Gigabit Ethernet cards in the six Win2K servers. Then, for each vendor, I replaced the existing Fast Ethernet switches with the vendor's Gigabit Ethernet equipment. I connected the client PCs to the desktop switch and used its 1000Base-T port to link it to the aggregation switch. Then, I connected the Win2K servers to the aggregation switch by using Cat 5 Enhanced (Cat 5e) patch cables for all the connections.

This test scenario might sound cut and dried, but I encountered many hurdles during my testing. You'll likely face similar problems when upgrading your networks, so I discuss them in the Web-exclusive sidebar "Upgrading to Gigabit Ethernet" (http://www.winnetmag.com, InstantDoc ID 24584). In general, I found that the right equipment with the right configuration can produce some worthwhile performance increases, although I never achieved anything close to true gigabit speed on the equipment I tested.

Two of the top-performing servers I tested were the Compaq ProLiant 7000 with a 64-bit 33MHz bus and the HP NetServer LT6000r with a 64-bit 66MHz bus. I used each server's high-performance SCSI RAID adapters in tandem with large arrays of Seagate Technology's Cheetah 10,000rpm SCSI drives. Intel PRO/1000 NICs were installed in each server, and I used Windows System Monitor to monitor throughput.

With some effort, I was able to coax the HP NetServer LT6000r to a maximum sustained network throughput of 460Mbps. I achieved this result during an application deployment test in which I pushed a 390MB Microsoft Office 2000 installation package to 18 Fast Ethernet clients while transferring similarly sized files to three other Gigabit Ethernet servers. To check general network responsiveness, I ran streaming multimedia content and used Windows Explorer to browse network resources on the two remaining Fast Ethernet clients. I didn't notice any image frame drop or jittering, and general network browsing was only slightly slower than usual. This test was simply impossible on Fast Ethernet equipment because it brought the network to a near standstill.

I noticed a significant improvement in multicast performance, measured by the time required to send a 690MB disk image to 18 multicast clients in one session. The HP NetServer LT6000r served as the multicast server, and the clients were using Fast Ethernet links to the desktop switch. On the Fast Ethernet network, the task took 19 minutes. On the Gigabit Ethernet network, the time was reduced to 9 minutes.

I measured the transfer of a large (1GB) file between the same hosts over Fast Ethernet and Gigabit Ethernet links with sustained network traffic (streaming media to multiple unicast clients). The file transfer took 230 seconds on Fast Ethernet and 88 seconds on Gigabit Ethernet.

Graph 1 and Graph 2 compare the peak performance numbers of Fast Ethernet and Gigabit Ethernet on three systems that I tested. The enterprise server realized a larger performance gain than the workstations because of higher I/O capacity. In essence, the Gigabit Ethernet equipment eliminated the network bottleneck and exposed the I/O bottleneck at the server or workstation.

Overall, my tests showed that Gigabit Ethernet provided a tangible performance improvement, but bottlenecks elsewhere kept the overall throughput lower than I had hoped. I was satisfied with Gigabit Ethernet performance relative to Fast Ethernet, and I was particularly impressed that general network responsiveness remained acceptable even during peak network loads. But I was disappointed not to be able to reach much beyond 450Mbps on the Lab's most capable server.

Although I saw tangible performance improvement on the Gigabit Ethernet servers, most networks will realize their greatest performance gain from Gigabit Ethernet over copper at aggregation points where multiple desktop switches bring hundreds of Fast Ethernet clients to converge on centralized server resources. Regardless of your deployment scenario, it's crucial that you do your homework beforehand so that you can forecast the performance gains you'll achieve in your particular environment. Then you can judge whether the gains will be worth the expense of upgrading.