IEEE

I first wrote about the potential issues with 10Gbps connectivity across a legacy fiber cable plant back in 2009 in an article titled, 10GBase-SR and Fiber Cable Plants. I just recently had another run into this same issue while working a consulting engagement. The customer had a legacy 62.5µm multimode fiber network that was currently running 1Gbps via 1000BaseSX GBIC transceivers. The customer was replacing their aging Cisco switches with HPE Aruba 2920 switches trying to use 10GBaseSR SFP+ transceivers. I was contracted to configure the HPE Aruba switches but when things didn’t work I had to dig through what the pre-sales team had assembled for the reseller and quickly found that someone had made some major mistakes in the design.

The customer believed that their fiber cable plant had 62.5µm multimode fiber and a spot check of the cable sheathing in a few locations confirmed as much. Unfortunately the fiber runs were believed to be between 200ft and 400ft in distance. With a 1000BaseSX transceiver you can reach 220m (721ft) over 62.5µm (OM1) fiber plant. With a 10GBaseSR transceiver you can only achieve 33m (108ft) over 62.5µm (OM1) fiber plant. That’s a pretty big difference and the primary reason why this initial upgrade attempt failed.

There is good news though for those with a legacy 62.5µm fiber cable plant, there’s no need to go tearing out that legacy fiber. The industry and the IEEE has provided a solution in 10GBaseLRM. The 10GBaseLRM transceiver can potentially reach 220m (721ft) over 62.5µm (OM1) fiber plant when combined with a special mode conditioning fiber patch cable. What’s a mode conditioning patch cable (MCPC)? It’s a special patch cable that has a small piece of singlemode fiber fusion spliced onto a multimode fiber.

ModeConditioningSetup — Image Credit: LANshack.com

Since singlemode fiber is usually only 9µm and multimode fiber can be either 50µm or 62.5µm the center of each fiber doesn’t match up when paired together, hence the mode conditioning fiber patch is required to help align the signal as it leaves the transceiver and hits the fiber cable plant.

The moral of this story, make sure you do the proper research and always verify that your fiber cable plant will be in spec for the transceiver you’re going to be using in your solution.

Cheers!

I was recently speaking with a colleague in Germany who was commenting about the recent request for public comment around the OUI Restructuring proposed by the IEEE RAC. You can find all the details in the draft document along with this presentation.

Unlike IPv4 addresses which are 32-bits long, the Ethernet MAC address is 48-bits in length and can provide a total of 281.5 trillion possible addresses. The first 3 bytes (24-bits) are reserved for identifying the vendor or manufacturer while the remaining 3 bytes (24-bits) are used to provide unique addresses to each device. As it exists today you could have 16.7 million unique addresses across 16.7 million unique vendors and/or manufacturers.

What devices have MAC address assigned;

Hardwired Network Adapters
Wireless Network Adapters
Bluetooth Adapters

You can find a MAC address in any of the following devices;

Smartphone
Tablet
Laptop/Desktop
Server
Virtual Server
Printer
Cable STB (Set-Top-Box)
Cable Router
Wireless Router
DVD/BlueRay player
Bluetooth Headset
IP Phone
IP Camera
Video Conference System
Switch/Router/Bridge
Wireless Access Point

The IEEE is proposing a change with how those addresses are allocated to help better utilize wasted address space as well as address virtualization challenges by creating large private address blocks for use within large virtualization deployments.

While a public IPv4 address needs to be unique across the entire Internet, a MAC address only needs to be unique across a Layer 2 network. It’s also worth noting that Layer 3 switches can have a unique MAC address for every port, so if you have a Layer 3 switch such as the Cisco Nexus 7010 or the Avaya VSP 9000 with 384 ports you’ll have 384 unique MAC addresses in that switch.

I downloaded the latest OUI table and counted about 17,597 assignments which means we have quite a ways to go before we exhaust the address space. I don’t see any issue with the proposed changes but I’m curious what everyone else thinks?

Cheers!

References;

http://en.wikipedia.org/wiki/MAC_address

10Gbps over 62.5µm Multimode Fiber

Are we running out of MAC addresses?

Recent Posts

Categories

SUBSCRIBE BY EMAIL