Shaping on Ethernet Links

[chira.cipri@gmail.com]

Hello community,

I have a question regarding class based shaping on Cisco routers. I am currently rereading Chapter 5 from the CCIE Routing&Switching Official Cert Guide Volume 2. This is the chapter on policing and shaping. 

The example used to describe traffic shaping, excerpt from page 211 in the aforementioned book. 

Routers can send bits out an interface only at the physical clock rate. To average sending
at a lower rate, the router has to alternate between sending packets and being silent. For
example, to average sending at a packet rate of half the physical link speed, the router
should send packets half of the time and not send packets the other half of the time.
Over time, it looks like a staccato series of sending and silence. Figure 5-1 shows a graph
of what happens when a router has a link with a clock rate of 128 kbps and a shaper configured
to shape traffic to 64 kbps.

 The example given is that of serial link of 128 kbps bandwidth shaped at 64kbps. 

Now, everyone knows that serial interfaces are indeed clocked, what with the CSU/DSU on serial interface and so on. My question is regarding shaping on GigabitEthernet interfaces.

Does the whole shaping mechanism of alternation between sending and silence in Tc interval still apply, or are there other methods to shape on Ethernet interfaces ? 

[Joseph W. Doherty]

It, QoS shaping, still applies, in the same manner, regardless of interface's media.

Also. BTW, a serial link's clocking has nothing to due with QoS shaping.

[chira.cipri@gmail.com]

---

@Joseph W. Doherty wrote:

It, QoS shaping, still applies, in the same manner, regardless of interface's media.

Also. BTW, a serial link's clocking has nothing to due with QoS shaping.

---

Well, this was in a way my question the physical clock rate still affects shaping as a whole. I was wondering if, either for ethernet or serial interfaces, on newer Cisco platforms, if there's away to tune shaping by modifying the underlying mechanics of the clock rate instead of just telling the interface "transmit at line rate for x seconds and then be silent for tc-x seconds." where tc=bc/CIR

[Joseph W. Doherty]

"Well, this was in a way my question the physical clock rate still affects shaping as a whole."

Yes and no.  On a serial interface, the clock rate effects the bit transmission rate, i.e. actual bandwidth.  The Ethernet "equivalent" would be 10 vs. 100 vs. 1,000 Mbps vs. etc.  We don't refer to either as "shaping".

"Shaping" is when we try to emulate a lower/slower transmission media, by using a medium less often then we physically can.

Of course you've read about the usual software shaping, but on least the older 3750 platforms, it could shape its four egress queues (and?/)or "shape" it's port.  The latter's description:

Bandwidth Limit Configuration:

In order to limit maximum output on a port, configure the srr-queue bandwidth limit interface configuration command. If you configure this command to 80 percent, the port is idle 20 percent of the time. The line rate drops to 80 percent of the connected speed. These values are not exact because the hardware adjusts the line rate in increments of six. This command is not available on a 10-Gigabit Ethernet interface.

srr-queue bandwidth limit weight1

where weight1 is the percentage of the port speed to which the port should be limited. The range is 10 to 90.

Notice, in the above, port is being idled for some percentage of time, actual transmission rate not being changed.

"Cisco platforms, if there's away to tune shaping by modifying the underlying mechanics of the clock rate instead of just telling the interface "transmit at line rate for x seconds and then be silent for tc-x seconds." where tc=bc/CIR"

I.e. for the 3750, above, clock rate not being changed, but yes, transmit for some time and hold for some time, although we don't configure Tc, Bc, CiR.