The Tolly Group recently put the D-Link DGS-1510 and DGS-3630 Managed Network Switches up against comparable switches from Cisco, HP Enterprise, and NETGEAR. The results of these tests are detailed here in this guest blog post from The Tolly Group founder, Kevin Tolly.


Overview

LAN switching is a key element of every business network – from the smallest office to high-performance data centers.  D-Link has products the run the gamut from budget-conscious layer 2 switches designed for small offices, to layer 3 Gigabit switches suitable for larger offices and enterprise access all the way through to 10 Gigabit Ethernet switches at the leading edge for high-performance networking.

The Tolly Group tested D-Link products at all three price points and compared them to popular switches from Cisco, HPE and NETGEAR.

The details of the tests and the full results can be found in three Tolly reports (noted below). In this blog, I will give you a high-level view of what we tested and what we found in The Tolly Group’s lab test.


Power Consumption: ATIS Weighted Average & TEER

LAN switches are “always-on” devices. Thus, differences in power consumption across switches can result in differences in operational costs. Furthermore, switches that consume more power will typically also require more cooling – increasing costs yet again.

The Alliance for Telecommunications Industry Solutions (ATIS) has developed several methods that assist in making “apples-to-apples” comparisons of switch power usage.

ATIS Weighted Average measures the power consumption of the switch under different traffic loads and provides an average consumption value that ATIS believes reflects common usuage.

ATIS Telecommunications Energy Efficiency Ratio (TEER) is a more recent standard that evaluates power consumption as it compares to throughput. TEER represents Gbps of throughput per Watt. With TEER, a higher number is better as it indicates more throughput for energy consumed.

The D-Link DGS-1510 and DGS-3630 performed very well in this category in recent testing.

Latency

Latency is delay – a measurement of time. With LAN switches, tests measure the time required for a packet to transit the switch. This measurement is important because delay accumulates as traffic travels, say, from a web server out to a client device. The goal is to add as little latency as possible as traffic travels through the network to its destination.

Fortunately, switches, in general, add very little delay. Switch latency, in fact, is measured in microseconds – or millionths of a second. And the switches tested add only a few microseconds of delay.

The D-Link DGS-1510 and DGS-3630 performed very well in this category in recent testing.


D-Link DGS-1510-28X Layer 3 Stackable Managed Gigabit Switch

This test compared the D-Link DGS-1510-28X Layer 3 Managed Gigabit Switch to similar switches from Cisco Systems, HPE and NETGEAR.  All of the switches tested provided at least 24 ports of Gigabit Ethernet and at least four ports of 10GbE. The full results of this test can be found in Tolly report #217116.

The switches tested were: D-Link DGS-1510-28X, Cisco SG500X-24, HPE OfficeConnect 1950-24G and NETGEAR S3300-28X.

The testing encompassed the following areas: Layer 2 & Layer 3 Performance (throughput & latency), MAC collision rates, and power consumption. Cost-per-Gigabit was also calculated.  Let’s look at the highlights of each of these areas.

Performance

Performance was tested with across 24GbE ports and four 10GbE ports and across the range of standard frame sizes (64-bytes through 1518-bytes) and run separately for both Layer 2 and Layer 3. The D-Link delivered wire-speed throughput at all payload sizes.  

Power Consumption

Tolly engineers used the standard ATIS weighted average to calculate power consumption. The D-Link switch had the lowest power consumption with an ATIS average of 17.  This was 51% better than Cisco and over 30% better than both HPE and NETGEAR.

MAC Address Collision Tests

Switches were tested up to their advertised MAC address table sizes of 16K entries. They were tested first using MAC addresses that were incremented and then tested again with randomly generated MAC addresses.

In the incremental MAC test, the D-Link switch missed only two addresses. This compared with 19 missed for Cisco, nine missed for HPE and 1,540 missed for NETGEAR.

With the random test, the D-Link switch missed 2,052 addresses. NETGEAR missed only 1,034 where Cisco missed 2,966 and HPE missed 2,054 addresses.

Cost-Per-Gigabit

Finally, engineers used the price and GbE throughput results to calculate cost-per-Gigabit of throughput. While throughput performance was the same for all of the switches, the price for the D-Link switch was the lowest, thus resulting in D-Link’s cost-per-Gigabit being the lowest. D-Link’s “cost per” was calculated as $20.42. This is 62% better than Cisco, 46% better than HPE and 21% better than NETGEAR.

In summary, the D-Link DGS-1510-28X matches the L2/L3 performance of the other switches tested, uses less power and provides the best cost/throughput value of all the switches tested.


D-Link DGS-3630-28TC Layer 3 Stackable Managed Gigabit Switch

This test compared the D-Link DGS-3630-28TC Layer 3 Managed Gigabit Switch to the Cisco Systems Catalyst 3650-24TD-E.  Both switches tested provided at least 24 ports of Gigabit Ethernet and at least two ports of 10GbE. The D-Link switch had 24GbE ports and four 10GbE ports and the Cisco switch had 26GbE ports but only two 10GbE ports. The full results of this test can be found in Tolly report #217112.

The testing encompassed the following areas: Layer 2 & Layer 3 Performance (throughput & latency), MAC collision rates, and power consumption. Cost-per-Gigabit was also calculated.  Let’s look at the highlights of each of these areas.

Performance

Performance was tested with across all GbE ports and all 10GbE ports and across the range of standard frame sizes (64-bytes through 1518-bytes) and run separately for both Layer 2 and Layer 3. The D-Link delivered wire-speed throughput at all payload sizes.  The D-Link switch also showed better (i.e., lower) latency than the Cisco switch in all test cases.

Power Consumption

Tolly engineers used the standard ATIS weighted average to calculate power consumption. The D-Link switch had better power consumption with an ATIS average of 33.07.  This was 38% better than Cisco.

MAC Address Collision Test

Switches were tested up to their advertised MAC address table sizes of 32,768 entries. They were tested first using MAC addresses that were incremented and then tested again with randomly generated MAC addresses.

Both switches were able to store all of the MAC addresses in the incremental test. With the random test, the Cisco Systems switch missed 91 addresses. The D-Link switch showed better results missing only 37 addresses.

Cost-Per-Gigabit

Finally, engineers used the price and GbE throughput results to calculate cost-per-Gigabit of throughput. While throughput performance was the same for all of the switches, the price for the D-Link switch was the lower, thus resulting in D-Link’s cost-per-Gigabit being the lower. D-Link’s “cost per” was calculated as $97.62. This is 53% better than Cisco, where cost-per-Gigabit was calculated as $220.54.

In summary, the D-Link DGS-3630-28TC matched the L2/L3 performance levels of the Cisco switch while delivering overall greater switch throughput with four 10GbE ports vs two 10GbE ports for Cisco. The D-Link switch uses less power and provides better cost/throughput value than the Cisco switch.


D-Link DXS-3400-24TC 10GbE Stackable Managed Switch

This test compared the D-Link DXS-3400-24TC 10GbEManaged Gigabit Switch to the NETGEAR M4300-24X. Both switches tested provided 24 ports of 10 Gigabit Ethernet. . The full results of this test can be found in Tolly report #217117.

The testing encompassed the following areas: Layer 2 Performance (throughput & latency), MAC collision rates, and power consumption. Let’s look at the highlights of each of these areas.

Performance

Performance was tested with across all 10GbE ports and across the range of standard frame sizes (64-bytes through 1518-bytes). The D-Link delivered wire-speed throughput at all payload sizes.  The D-Link switch also showed the same latency as the NETGEAR switch in all test cases.

Power Consumption

Tolly engineers used the standard ATIS weighted average to calculate power consumption. The D-Link switch had an ATIS rated average of 114.52 compared to 104.64 for NETGEAR.

MAC Address Collision Test

The NETGEAR switch specification states a MAC table size of 16K addresses and it was tested at that level. The D-Link switch specification states a MAC table size of 48K and it was tested with 32K addresses. The switches were tested first using MAC addresses that were incremented and then tested again with randomly generated MAC addresses.

In the incremental MAC test, the D-Link switch did not miss any addresses. This compared with 1,671 missed addresses for NETGEAR.

With the random test, the D-Link switch missed 37 addresses. NETGEAR missed 1,950 addresses

In summary, the D-Link DXS-3400-24TC matched the L2 10GbE performance levels – throughput and latency – of the NETGEAR switch while providing better MAC address collision test results and similar power consumption.