.shock - Fotolia

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

Advent of 2.5, 5 Gigabit Ethernet switches roiling market

Faster wireless has opened the market for a new breed of switches. Expert Craig Mathias will tell you what you need to know.

With a number of recent announcements unveiling switches supporting per-port speeds of 2.5 and 5 Gigabit Ethernet (GbE), it's time to look at what impact this very interesting advance might have on the networking market -- both wired and wireless. It's actually Gigabit+-class wireless LANs -- based on 802.11ac -- that are driving interest and demand here; after all, many current WLAN products can easily exceed the 1 GbE per-port speeds available on most Ethernet switches deployed at the network's wired edge.

The next jumping-off point after 1 Gbps had been 10 Gbps, but the (minimum) requirement for Category 6 cable here means buildings wired with Cat 5e can't take advantage of 10 Gbps -- at least not at full-rated performance of throughput and distance. And 10 GbE switches have historically been much more expensive than 1 GbE models, again on a per-port basis.

This is why 2.5 and 5 GbE technology is so interesting. It provisions performance that matches the capabilities of today’s 802.11ac access points (APs), but over Cat 5e cabling. This means a boost in performance without the need to change or replace wiring. Yet, as is often the case, a new set of considerations has materialized among them:

  • While new cabling may not be required, new switches obviously are. To that end, IT managers need to evaluate price/performance and determine the net benefits of adding or replacing these switches. They also need a budget to proceed with such an upgrade, which almost always involves looking a few years into the future along with prioritizing needs based on quantified benefits. What's important here is the useful life of 2.5 and 5 GbE: Will 10 GbE eventually be required regardless? Read on.
  • The opportunity to get additional throughput without the need to upgrade wiring is tempting, but there's more to think about here. All newly installed cabling will be at least Cat 6 today, but it might be easier to replace cabling than many are aware: Just tape the new cable to the old and pull. OK, sometimes this won't be easy, but it's likely not every link will need to be upgraded. Indeed, 10 GbE will often work quite well over Cat 5e cabling at shorter distances; in our experience, this can be up to 50 meters. Given that APs are often densely deployed, this option should be acceptable in many cases.
  • Another tradeoff involves the often much-higher cost of 10 GbE switches, again on a per-port basis. But the appropriate question here is really how fast 10 GbE prices will decline -- as prices of networking equipment always do. We believe that 10 GbE pricing will fall quite rapidly indeed. In addition, we believe 10 GbE performance will ultimately be required. That said, the cost of entry remains a consideration today. Organizations should perform an analysis of per-port prices of 2.5 and 5 Gigabit Ethernet switches versus their 10 GbE counterparts -- ideally through a competitive-bidding process. The difference might already be manageable, if not negligible.
  • This brings us to the most important question of all: how to address the next generation of WLANs. The 802.11ad protocol can reach close to 7 Gbps and .11ax, now under development at the IEEE and due around 2018, is aiming at the magic 10 Gbps over-the-air number. Wave 2 Gigabit Wi-Fi, meantime, should be fine with 5 Gbps links; even though 802.11ac is specified at up to 6.7 Gbps. It's unlikely these products will become common, partially because they involve the use of eight multiple-input multiple-output streams, which are very difficult to implement in mobile devices (remember, that's eight radios and eight antennas -- a logistical and power challenge if there ever was one). But dual-band Wave 2 APs will surpass 2.5 Gbps in many cases; we already have such devices in operation in our lab, and some implementations could strain even 5 Gbps.
The opportunity to get additional throughput without the need to upgrade wiring is tempting, but there's more to think about here.

While all of this is being sorted, keep in mind that standards for 2.5 and 5 GbE are currently under development, with at least two major camps -- NBASE-T and MGBASE-T-- competing. We don't, however, believe that there is much risk in purchasing these products today, as these will almost certainly be forward-compatible with any final standard. But it's worth noting that standards for 10 GbE have been around for some time.

Quite a dilemma, indeed. What's clear is that all IT shops are going to need, at some point, to upgrade their Ethernet switch installed base in order to get the most out of contemporary and future wireless LANs. But 2.5 and 5 GbE, viewed as an upper performance boundary in a given product, could be asking for trouble in the form of yet another partial upgrade cycle required with two or three years.

Our recommendation, then, is to look for switches that today offer 1, 2.5, 5 and 10 GbE on every port, equipped with power over Ethernet (PoE) as required. Such a strategy is essentially future-proof. It's very unlikely that anyone's planning horizon could consider wireless LANs operating above 10 Gbps -- these are unlikely to appear in the next 10 years and perhaps beyond that (which is why this is a topic for another day). Switches offering this combination of port speeds will ultimately yield the best value for end-user organizations, both large and small, for many years to come.

Next Steps

Getting ready for Gigabit speeds

Taking your wired network to the next step

How to buy your next access point

This was last published in March 2015

Dig Deeper on Network Infrastructure