<s>
High-availability	B-Protocol
Seamless	I-Protocol
Redundancy	I-Protocol
(	O
HSR	O
)	O
is	O
a	O
network	O
protocol	O
for	O
Ethernet	O
that	O
provides	O
seamless	O
failover	B-General_Concept
against	O
failure	O
of	O
any	O
single	O
network	O
component	O
.	O
</s>
<s>
PRP	O
and	O
HSR	O
are	O
independent	O
of	O
the	O
application-protocol	O
and	O
can	O
be	O
used	O
by	O
most	O
Industrial	B-Protocol
Ethernet	I-Protocol
protocols	O
in	O
the	O
IEC	O
61784	O
suite	O
.	O
</s>
<s>
HSR	O
nodes	O
have	O
two	O
ports	O
and	O
act	O
as	O
a	O
bridge	B-Protocol
,	O
which	O
allows	O
arranging	O
them	O
into	O
a	O
ring	O
or	O
meshed	O
structure	O
without	O
dedicated	O
switches	O
.	O
</s>
<s>
This	O
is	O
in	O
contrast	O
to	O
the	O
companion	O
standard	O
Parallel	B-Protocol
Redundancy	I-Protocol
Protocol	I-Protocol
(	O
PRP	O
)	O
,	O
with	O
which	O
HSR	O
shares	O
the	O
operating	O
principle	O
.	O
</s>
<s>
For	O
such	O
applications	O
,	O
the	O
recovery	O
time	O
of	O
commonly	O
used	O
protocols	O
such	O
as	O
the	O
Rapid	O
Spanning	B-Protocol
Tree	I-Protocol
Protocol	I-Protocol
(	O
RSTP	O
)	O
is	O
too	O
long	O
.	O
</s>
<s>
The	O
cost	O
of	O
HSR	O
is	O
that	O
nodes	O
require	O
hardware	O
support	O
(	O
FPGA	B-Architecture
or	O
ASIC	O
)	O
to	O
forward	O
or	O
discard	O
frames	O
within	O
microseconds	O
.	O
</s>
<s>
An	O
HSR	O
network	B-Protocol
node	I-Protocol
(	O
DANH	O
)	O
has	O
at	O
least	O
two	O
Ethernet	O
ports	O
,	O
each	O
attached	O
to	O
a	O
neighbour	O
HSR	O
node	O
,	O
so	O
that	O
always	O
two	O
paths	O
exist	O
between	O
two	O
nodes	O
.	O
</s>
<s>
Therefore	O
,	O
as	O
long	O
as	O
one	O
path	O
is	O
operational	O
,	O
the	O
destination	O
application	O
always	O
receives	O
one	O
frame	B-Protocol
.	O
</s>
<s>
HSR	O
is	O
typically	O
used	O
in	O
a	O
ring	B-Architecture
topology	I-Architecture
or	O
in	O
another	O
mesh	B-Architecture
topology	I-Architecture
.	O
</s>
<s>
Redundant	O
connections	O
to	O
other	O
networks	O
are	O
possible	O
,	O
especially	O
to	O
a	O
Parallel	B-Protocol
Redundancy	I-Protocol
Protocol	I-Protocol
(	O
PRP	O
)	O
network	O
.	O
</s>
<s>
it	O
can	O
forward	O
a	O
frame	B-Protocol
received	O
on	O
one	O
port	O
to	O
at	O
least	O
one	O
other	O
port	O
in	O
cut-through	O
mode	O
.	O
</s>
<s>
A	O
source	O
node	O
sends	O
the	O
same	O
frame	B-Protocol
over	O
all	O
ports	O
to	O
the	O
neighbour	O
nodes	O
.	O
</s>
<s>
A	O
destination	O
node	O
should	O
receive	O
,	O
in	O
the	O
fault-free	O
state	O
,	O
two	O
identical	O
frames	O
within	O
a	O
certain	O
time	O
skew	O
,	O
forward	O
the	O
first	O
frame	B-Protocol
to	O
the	O
application	O
and	O
discard	O
the	O
second	O
frame	B-Protocol
when	O
(	O
and	O
if	O
)	O
it	O
comes	O
.	O
</s>
<s>
A	O
node	O
forwards	O
a	O
frame	B-Protocol
unless	O
it	O
detects	O
a	O
frame	B-Protocol
that	O
it	O
sent	O
itself	O
or	O
that	O
it	O
already	O
sent	O
.	O
</s>
<s>
To	O
reduce	O
traffic	O
,	O
a	O
node	O
may	O
refrain	O
from	O
sending	O
a	O
frame	B-Protocol
that	O
it	O
already	O
received	O
from	O
the	O
opposite	O
direction	O
on	O
the	O
same	O
port	O
(	O
Mode	O
X	O
)	O
,	O
but	O
this	O
does	O
not	O
apply	O
to	O
all	O
frames	O
.	O
</s>
<s>
Also	O
,	O
several	O
algorithms	O
that	O
relies	O
on	O
network	B-Protocol
node	I-Protocol
location	O
learning	O
can	O
serve	O
in	O
the	O
HSR	O
traffic	O
reduction	O
,	O
such	O
as	O
the	O
Port	O
Locking	O
and	O
Enhanced	O
Port	O
Locking	O
,	O
(	O
PL	O
)	O
and	O
(	O
EPL	O
)	O
respectively	O
,	O
which	O
work	O
on	O
closing	O
the	O
ports	O
that	O
leads	O
to	O
a	O
non	O
existed	O
node	O
,	O
</s>
<s>
Especially	O
,	O
Precision	B-Protocol
Time	I-Protocol
Protocol	I-Protocol
frames	O
(	O
multicast	O
)	O
are	O
no	O
duplicates	O
of	O
each	O
other	O
since	O
they	O
are	O
modified	O
by	O
each	O
node	O
to	O
correct	O
the	O
time	O
.	O
</s>
<s>
A	O
special	O
treatment	O
is	O
given	O
to	O
link-specific	O
frames	O
such	O
as	O
LLDP	B-Protocol
or	O
Pdelay_Req	O
/	O
Pdelay_Resp	O
Precision	B-Protocol
Time	I-Protocol
Protocol	I-Protocol
frames	O
,	O
for	O
which	O
the	O
HSR	O
tag	O
is	O
ignored	O
,	O
but	O
must	O
be	O
present	O
.	O
</s>
<s>
To	O
simplify	O
the	O
detection	O
of	O
duplicates	O
,	O
the	O
frames	O
are	O
identified	O
by	O
their	O
source	O
address	O
and	O
a	O
sequence	O
number	O
that	O
is	O
incremented	O
for	O
each	O
frame	B-Protocol
sent	O
according	O
to	O
the	O
HSR	O
protocol	O
.	O
</s>
<s>
The	O
sequence	O
number	O
,	O
the	O
frame	B-Protocol
size	O
and	O
the	O
path	O
identifier	O
are	O
appended	O
in	O
a	O
6-octet	O
HSR	O
tag	O
(	O
header	O
)	O
.	O
</s>
<s>
NOTE	O
:	O
all	O
legacy	O
devices	O
should	O
accept	O
Ethernet	B-Protocol
frames	I-Protocol
up	O
to	O
1528	O
octets	O
,	O
this	O
is	O
below	O
the	O
theoretical	O
limit	O
of	O
1535	O
octets	O
.	O
</s>
<s>
However	O
,	O
since	O
also	O
the	O
network	O
infrastructure	O
is	O
also	O
doubled	O
in	O
closed	O
ring	B-Architecture
topologies	I-Architecture
the	O
nominal	O
network	O
bandwidth	O
is	O
available	O
.	O
</s>
<s>
is	O
required	O
to	O
bring	O
down	O
the	O
per-hop	O
latency	O
to	O
a	O
reasonable	O
value	O
(	O
some	O
5μs	O
at	O
100	O
Mbit/s	O
)	O
,	O
using	O
cut-through	B-Protocol
switching	I-Protocol
.	O
</s>
<s>
To	O
this	O
purpose	O
,	O
each	O
frame	B-Protocol
has	O
an	O
HSR	O
tag	O
that	O
allows	O
recognition	O
of	O
whether	O
the	O
frame	B-Protocol
should	O
be	O
forwarded	O
or	O
not	O
,	O
to	O
avoid	O
store-and-forward	B-Protocol
.	O
</s>
<s>
IEC	O
62439-3	O
Annex	O
C	O
specifies	O
a	O
Precision	B-Protocol
Time	I-Protocol
Protocol	I-Protocol
Industry	I-Protocol
Profile	I-Protocol
(	O
PIP	O
L2P2P	O
)	O
,	O
that	O
allows	O
a	O
clock	O
synchronization	O
down	O
to	O
an	O
accuracy	O
of	O
1	O
μs	O
in	O
a	O
ring	O
of	O
16	O
HSR	O
nodes	O
.	O
</s>
<s>
It	O
has	O
been	O
adopted	O
by	O
IEEE	O
as	O
IEC/IEEE	B-Protocol
61850-9-3	I-Protocol
,	O
</s>
