[j-nsp] Accuracy of interface stats

[Stefan Fouant]

On Sun, Sep 27, 2009 at 4:03 AM, Christoph Blecker <cblecker at peer1.com>wrote:

> Additionally, if you're running it as OC-192 with SONET, as opposed to
> 10GbE, the saturation point is lower due to the additional overhead of
> the SONET protocol. It's like the theoretical of an OC48 is 2.488Gbps,
> but you can experience saturation as low as 2.2-2.3Gbps.
>
I'm curious how assumptions are being made that somehow SONET overhead is
larger than Ethernet overhead.  I think it largely depends on the frame size
being transmitted.  Section, Line, and Path overhead typically equate to
about 3.3% of the SONET frame, which on a SONET OC-192 interface leaves
roughly 9.67 Gbps for payload.  If transporting 64 Byte IP packets and
factoring in 10 bytes of overhead per packet (assuming RFC 2615 PoS framing
utilizing PPP over HDLC with a 4 Byte FCS), we can transmit roughly
16,334,459 frames per second:

64 Byte IP packet + 10 Byte PPP Overhead = 74 Bytes * 8 bits per byte = 592
bits per frame

9,670,000,000 bps available payload / 592 bits per frame = 16,334,459 fps

If we take a look at 64 Byte IP packets on 10GigE on the other hand, using
standard IEEE 802.3 framing, we get considerable less throughput:

64 Byte IP packet + 8 Byte Ethernet Preamble + 18 Byte Ethernet header = 90
Bytes * 8 bits per byte = 720 bits per frame

10,000,000,000 bps GigE / 720 bits per frame = 13,888,888 fps

Have my calculations been incorrect all these years?  What am I missing?

--
Stefan Fouant