[Noisebridge-discuss] what could we do with one of these babies..
joel jaeggli
joelja at bogus.com
Sun Apr 18 21:12:12 UTC 2010
On 4/18/2010 1:25 PM, Jesse Zbikowski wrote:
> On Sat, Apr 17, 2010 at 8:47 PM, Joel Jaeggli<joelja at bogus.com> wrote:
>> On 04/17/2010 06:09 PM, Jesse Zbikowski wrote:
>>> EtherCAT is a hack on Ethernet which allows you to address 100's of
>>> devices in microseconds (as opposed to milliseconds).
>>
>> Well, the normal lower latency bound for ethernet is the size of the
>> packet at a given rate and the latency associated with the serdes if
>> there is one in the path. at 10Gbe rates, a single 1500byte pdu can be
>> clocked out at a rate of every 1.2us, which goes up to 7us at 9k frame size.
>
> Right, the transmission latency of a single Ethernet packet is very
> low. This is what makes it a good basis for EtherCAT. Indeed in the
> case of a single transmitter and a single receiver, I expect the
> advantages of EtherCAT would be moot.
>
> Once there are multiple senders and receivers on the bus, the story
> changes. With standard Ethernet if two senders talk at the same time,
> you have a packet collision. This forces them to wait for a random
> amount of time, retransmit, and repeat -- as many times as necessary.
> So the worst case is theoretically unbounded badness; you could be
> sitting around all day waiting to talk, or at least milliseconds.
half duplex gigabit ethernet is rare enough that i would hazard you've
never used a system that was both gigabit and used a repeater (hub),
there is no half duplex 10Gig,
> EtherCAT on the other hand uses a ring topology, where every device
> talks in turn and there is no bus contention.
The realtime process control industry loved arcnet for precisely this
reason, on the other hand it was also slow and pretty muck failed at
both the standardization and the commiditization exercise, which is why
this has pretty much all moved over to ethernet.
> Hence our jitter (delay
> between when we expect a signal and when we actually get it) is low;
> less than a microsecond. Deterministic behavior is the key to
> realtime.
right, but by moving forward a couple of generations on the technology
curve you can move 1 or 2 orders of magnitude more data with an order of
magnitude lower latency which doesn't sound that painful.
> Certainly there are systems which can use normal Ethernet, especially
> if the number of devices is low and realtime requirements are soft.
> Determinism is hard to fake though.
>
> http://en.wikipedia.org/wiki/Ethercat#Synchronization
ieee 1588 works just fine over regular ethernet, I personally am not a
fan of time syncronization at the link layer unless you need clock
recovery for framing, (this is ethernet so you don't, contrast with SDH
which is 4k superframes per second regardless of speed) it has rather
ugly implications what happens when intermediary l3 structures are
imposed on what was previously a single broadcast domain.
More information about the Noisebridge-discuss
mailing list