Clock and data recovery is an essential physical-layer function of modern switch and router hardware. Digging deep into the electronics of a router may not be your thing, but clock recovery is a fundamental building block for other network hardware functions. For example, serial to parallel data conversions require reliable clock and data recovery (CDR) to function effectively. It’s hard to understand serial to parallel conversions or signal conditioning without learning about CDR first.
Modern top-of-rack switches (or TORs) run at line rate and are non-oversubscribed. This means you get non-blocking  port-to-port throughput within the switch ASIC at the line rate of the front panel ports. Almost all TOR switches use a single switch ASIC and the industry demanded port-density on a single ASIC, and the manufacturers delivered. The list below shows the 10Gbps port density evolution of the Broadcom StrataXGS product line. The Intel Fulcrum ASIC evolution isn’t shown here but looks very similar.
- Scorpion: 24 x 10Gbps
- Trident: 48 x 10Gbps
- Trident+: 64 x 10Gbps
- Trident2: 108 x 10Gbps (108 x 10G MACs – can handle 1.2Tbps using some ports at 40G). Continue reading
If you need regular console port access then nothing beats a fixed console router. However there are many times when that simply isn’t an option. For occasional console connections I use a Keyspan USB/Serial adaptor with my MacBook. It’s an acceptable solution but not sometimes is far from ideal. You can find yourself tethered to the rack in question, often sitting on the floor or a crappy stool, dealing with datacenter noise and temps and have console cables traipsed across the aisles. Surely there’s a better way to do this?
Over the years I’ve seen the eye pattern diagram a few times. At first I don’t understand it and I get an instructor or peer to explain it. I briefly understand, then moments later… it’s gone. If you grok how this diagram is built straight away then please forgive me. Otherwise please read on.
The eye pattern diagram is named after the eye shape in the middle of the diagram above. The purpose of the diagram is to measure the quality of the received electrical or optical signal. Put simply, the bigger the eye opening the better the signal. Before we dig deeper on the measurement we’ll examine how the eye diagram is made.
Every now and again you see a snippet of complex CLI syntax that gives you pause for thought. Last week I saw the command below in a change procedure. The command was being used to verify baseline BGP neighbor state and re-verify after a policy change.
1 2 3
show ip bgp peer-template eBGP_Peers | egrep default | sed 's/default://' \ | tr -s ' |\n' | tr -s ' ' '\n' | sed 's/^/show ip bgp nei /' \ | sed 's/$/ adver | grep \//' | vsh
I was a little daunted by the complexity of the command at first. I was slightly embarrassed too, as I had no idea what the command did. I learned that this command finds all neighbors which use a named bgp peer-template, and lists the prefixes they advertise. In this post I’ll break down the command and share the love about NXOS CLI and bash scripting. Continue reading
We commonly interconnect network devices over distances of 100 meters for Cat5 Ethernet or maybe 300 meters over OM3 multimode fiber for 10GBASE-SR. With enough money we can extend the span between devices to perhaps 10Km for with 10GBASE-LR), 40Km for 10GBase-ER or even 80Km for 10GBase-ZR.
Why then, when we move to the world of circuit boards, do we have to suffer maximum distances which are measured in hundreds of millimeters? Let’s explore why life is so hard for the high-speed signals which transit the circuit board. Continue reading