This document answers frequently asked questions in regards to Spatial
Reuse Protocol (SRP) and Dynamic Packet Transport (DPT) Cisco hardware and
Where can I find the DPT feature guide?
A. Refer to the
Reuse Protocol Feature Guide in order to find the DPT feature
Can DPT carry 802.1q frames?
A. With the Cisco 10720 router, with Universal Transport Interface (UTI)
support, and the tunnel server card on the Gigabit Switch Router (GSR), you are
able to take Ethernet frames, and encapsulate the frames to the UTI. You can
then carry the encapsulated frames over the DPT ring, and to the GSR tunnel
server card in order to process.
How do I measure a new DPT ring segment for quality and
A. These Cisco IOS® Software
debug commands can be used in order to check Layer 2
(L2) protocols, once a ring is brought up:
Send four types of traffic and issue the show interface
srp and show srp counters commands in
order to check these counters:
Unicast low-priority traffic (default Type of Service (ToS) 0 to
Unicast high-priority traffic (default ToS 6 to 7). Be careful of the
default 20mB rate limiter.
Multicast low-priority traffic (default ToS 0 to 5)
Multicast high-priority traffic (default ToS 6 to
With regard to the Bit Error Rate (BER), this information
You can read the BER for B1, B2, and B3 from the output of the
show controller command.
You can change the thresholds for B1, B2, and B3 the same way you can
for a normal Packet over SONET (PoS) link.
You cannot see any BER counts in the ring unless there is an
extremely long haul, for example 70 to 80 km or more.
The range for the BER threshold is -3 to -9, though you cannot see
any B1, B2, or B3 errors in a well-built
For specific SRP and DPT equipment, refer to
, who offer SRP and
DPT test equipment. You can tell if the line card is operational, whether
messages are exchanged, with these products. The Spirent (Adtech) system can
create messages in order to simulate an operational ring (Intelligent
Protection Switching (IPS), keep alives, and topology). Both of these products
are software extensions to their OC-48 PoS testers.
What is the overhead created by DPT to an IP packet?
A. SRP overhead is 21 bytes above the base IP packet, which is 16 bytes
OH, 4 bytes Frame Check Sequence (FCS) and 1 byte delimiter. The data usage is
minimal for control packets. There are packets for IPS, topology, node name,
and usage, which depends on the configuration. This totals approximately 2000
packets per second, which is mostly usage. All of these are small packet sizes
(40 to 128 bytes), which amount to about 0.05 percent of the traffic.
How do you configure SRP MAC accounting?
A. Issue these commands in order to configure SRP MAC accounting:
Issue the show srp source-counters command
as shown in this example in order to view the results:
srp-router#show srp source-counters
Source address information for interface SRP0/0 is shown in this
What is the benefit to run DPT over SONET with a protected or unprotected
Benefits of DPT over SONET
A. The main benefit of running DPT over SONET is the fact that you use a
technology that is optimized to carry IP or data traffic while you maintain the
existing Time-Division Multiplexing (TDM) services. This way you introduce
statistical multiplexing onto a TDM infrastructure. All of this is over a
DPT over SONET with a Bidirectional Line Switched Ring (BLSR) or a Unidirectional Path Switched Ring (UPSR)
If you run DPT over unidirectional path switched ring (UPSR), the only
practical way is to run this over an unprotected UPSR. A device such as the
Cisco ONS 15454 offers this capability, but not all Add Drop Multiplexers
(ADMs) do. In this situation, you must rely on the DPT protection in the case
of failures. If there is a failure, the DPT protection, Intelligent Protection
Switching [IPS], takes affect and you have a wrapped DPT ring.
In the case of DPT over bidirectional line switched ring (BLSR), if
there is a failure, the BLSR protection kicks in and you have no wrap in the
DPT ring. This means more bandwidth at all times. The only time DPT protection
is activated is in the case of a failure between the DPT router and the ADM.
You cannot create unprotected SONET circuits over a BLSR ring. BLSR uses shared
protection and assumes that every circuit uses this protection.
Does the OC-12 DPT line card (Engine 1) implement high- and low-priority
transit and transmit queues for the SRP-FA?
A. The OC-12 DPT line card has only one queue in the transmit path, and
two queues in the transit path. However, the rings operate on a single-queue
basis due to the single transmit queue.
The SRP-Fairness Algorithm (FA) only works on the low-priority queue
(which is implemented) and never operates on the high-priority queue. There is
no low- or hi-rate limiting on the OC-12 DPT line card.
In addition, the four-port OC-12c/STM-4c DPT Internet Service Engine
(ISE) line card, Cisco 12000 and 12400 series is based on Engine 3. This line
card fully supports hi and low SRP queues and full modular Quality of Service
(QoS) Command Line Interface (CLI) (MQC). The customer is able to change the
priority slicing and assign specific types of packets to a specific queue. The
line card also allows any traffic policy to assign any action, such as
bandwidth or Type of Service (ToS) changes.
Note: Refer to
Software: Quality Of Service for more information on QoS.
How many nodes can a DPT ring accommodate?
A. For an STM-16 DPT ring, this information applies:
You are limited to 62 node rings if you use the older Frame Check
Sequence (FCS) version of DPT (rev-A). This is also true if you mix rev-A and
rev-B versions of the DPT card.
The new limit is 128 node rings, if all your nodes use the newer
For an STM-4 DPT ring, this information applies:
Is SRP or DPT the correct term to use?
A. Cisco DPT is the type of network architecture customers can build,
based on the Cisco SRP MAC architecture and protocol. In the future customers
are able to build Resilient Packet Ring (RPR) network architecture, based on
the IEEE 802.17 MAC architecture and protocol. DPT/RPR is the naming the market
and customers use.
These are definitions of the terms mentioned:
RPR—The name of the category of products and technologies that
deliver RPR functionality.
DPT—The product line name for the Cisco family of RPR products, such
as the OC-48 DPT line card for the Cisco 12000 series router.
SRP—The name of the Cisco-developed MAC-layer protocol and the
underlying technology used in the Cisco DPT and RPR family of products. SRP is
an open, freely available, specification (RFC 2892
), and has
been submitted to the IEEE for consideration as the basis of the forthcoming
802 standard MAC-layer implementation.
IEEE 802.17—The name of the forthcoming standard MAC-layer protocol
implementation for an RPR.
Can a Gigabit Switch Router (GSR) OC-48 DPT card be downgraded to an
A. No, this is not possible. There are two areas that limit this
capability. This is the DPT stack:
DPT/SRP RAC ASIC <--> SONET/SDH framer <--> Optics PHY
The Resource Availability Confirmation (RAC) Application-Specific
Integrated Circuit (ASIC) for OC-12 is a version 1 Spatial Reuse Protocol (SRP)
ASIC. The RAC ASIC for OC-48 is a version 2 SRP ASIC. There are a few small
differences between version 1 and 2. Both run their own fixed ASIC clock
Both framers, for OC-12 and OC-48, run their own fixed framer clock
rate. A framer supports one interface line
Can you mate a C48/SRP-SR (Short Reach line card) and OC48/SRP-LR
(long-reach line card) in a Gigabit Switch Router (GSR)?
A. There are no issues if you mix SR and LR OC-48s with SRP in the same
GSR. This has been extensively tested, and there are no restrictions. The only
concern is if an SR or LR is fiber-connected to a line card with a different
reach, such as a SR line card connected to a LR line card over fiber. In this
case, you must use attenuation in order to bring power levels down in the
Can you provide information on SRP bandwidth?
A. SONET line rate (for an OC-48) is 2488.32 Mbps. The overhead quick
calculation is 1 byte per 27 bytes transmitted. Therefore, the available
payload is approximately 26/27 or 2488.32 = 2396.16 Mpbs.
The number that is usually used for general calculations, for rough
math, is 2.395 Gbps. This number takes into account Path OverHead (POH)). This
is the bandwidth available in order to insert SRP control packets and data
You always have the full 2.395 available to the SRP, and while SRP
control packets take up almost no bandwidth (even keep alive at 106us intervals
is almost nothing), the size of packets with 16-byte SRP overhead can make a
big difference to your IP bandwidth. For example, 40-byte IP packet = 56 byte
SRP packet = 40/56 * 2.395 = 1.71 Gbps of IP traffic even though SRP uses all
2.395 G. However, a 1500-byte IP packet = 1516 byte SRP packet = 1500/1516 *
2.395 = 2.369 Gbps of IP traffic even though SRP uses all 2.395 G.
What is Single Ring Recovery (SRR)?
A. SRR deals with multiple fiber failures on a single ring. The SRR
protocol allows DPT to run over a single ring when two or more failures are on
the same ring. The SSR protocol enables an SRP ring to preserve full-node
connectivity in the event of multiple failures on one of its two
counter-rotating rings (Inner Ring (IR) or Outer Ring (OR)), while the other
ring is failure free. In all other cases, such as dual ring failures, the SRP
ring maintains the standard SRP Intelligent Protection Switching (IPS)
These are the rules:
If it is a single failure, use IPS.
If there are multiple failures at the same ring, each node initiates
SRR is an extension to the SRP. SRR includes these two new SRP control
These allow for each router to learn about the failures in the ring.
Discovery packets are sent every ten seconds when enabled on all ring nodes. If
a ring node detects a local failure, the node launches a discovery packet on
both rings. Every ring transit node updates the packet with its own failure
information. The originator launches an announce packet that indicates the
number of failures on each ring when the topology discovery packet
Note: Topology packets are sent point-to-point to MAC address
Also, the SRP fairness algorithm does not work when a single ring is
used. The bandwidth of each node is hard limited, and the per-node bandwidth
limit is 100M with OC-12/STM-4 and 400M with OC-48/STM-16. SRR is a software
release implementation and is not enabled by default. The show srp
srr command reports the status of the SRR feature. Refer to
Single Ring Recovery
Protocol for more information.
How does the 1310nm laser signal interconnect with a 1550 nm laser
A. A 1550 nm laser signal, at a 1550 nm interface, can be received by or
detected by a diode at the 1310 nm interface. A 1310 nm laser signal, at a 1310
nm interface, can be received by or detected by a diode at the 1550 nm
The reason for this is that all the optical router interfaces, DPT and
Packet over SONET (PoS), use the receive (Rx) part of the interface (a wideband
diode). This means the diode can receive either 1310 nm or 1550 nm laser
In general, you can use the rules in this section as a guideline for a
STM-16 long distance dark fiber design. This example is based on the Long Reach
2 (LR2) interface. But, similar rules apply for the Long Reach 1 (LR1)
interface. The dispersion is less of an issue with the 40 km fiber. The fiber
attenuation at 1310 nm, used with the LR1 interface, is higher.
This is an example with an STM-16 LR2.
There are two parameters that are important in a long distance dark
Fiber media specifications with respect to loss (dB/km at 1550 nm) and
dispersion (ps/nm/km) are critical at these distances.
Too many or too few amplification and dispersion limitations generate
ring wrap conditions due to a signal degrade condition. This is indicated in
the output from the show controllers srp command.
This is usually due to improper optical power levels or high dispersion levels.
These are two critical parameters in such a long network span. Too high or too
low power, with edge value conditions, can also cause a lot of bit
G.652 and G.653, or fiber with similar specifications, are two commonly
used fiber types. Regular G.652 Single-Mode Fiber (SMF) is optimized for zero
dispersion around 1310 nm. This is not optimal for 1550 nm transmission, used
with a LR2 interface. Therefore, G.653 DS was developed with zero dispersion at
Common fiber loss examples are 0.2 to 0.4 dB/km at 1550 nm. About 0.30
dB/km for dark fiber is middle-class quality fiber. This does not include any
span or segment interconnect loss.
The LR2 PHY is tested in order to ensure that is is less than the
International Telecommunication Union (ITU) mandated optical path penalty.
Vendor specification of the LR2 optics is characterized to 1800 ps/nm of the
total dispersion. As an example, the maximum span can be 100 km at the limit of
dispersion tolerance, in the case of an 18 ps/nm/km fiber.
These are the specifications for the SMF LR2 interface:
Operating wavelength 1550 nm
Transmit power 3 dBm (max) -2 dBm (min)
Receive sensitivity -9 dBm (max) -28 dBm (min)
Recommended distance 80 km
Power budget 26 dB
You need to calculate for a worse-case scenario. This can include
connector loss, splices, aging of optics, aging of fiber, and patch cords,
which could be 3 to 4 dB in total. Such a cable is usually laid down in
segments, and the interconnections also take up some of the budget.
The maximum span is approximately 86 km with a power budget of 26 dB
and a fiber attenuation per km of 0,3 dB. For example, in the case of a 23 dB
power availability (26 - 3 = 23), the maximum span can be 76 km at the limit of
The maximum span is approximately 104 km with a power budget of 26 dB
and a fiber attenuation per km of 0,25 dB. As an example, in the case of a 23
dB power availability (26 - 3 = 23), the maximum span may be 92 km at the limit
of power tolerance.
Both of these examples show that there is a certain delta, and the
fiber media specifications and additional loss matter. The LR2 80 km
recommended distance is just a save value. You never work with these fixed
numbers in optical networking, in general. This is because there are too many
variable optical parameters involved.
Real loss measurement, or fiber media vendor specifications, is a
requirement in order to design dark fiber-based DPT and Resilient Packet Ring
In case a span is more than 80 km, the 15104 can be considered as a 3-R
regenerator. The 15104 has only LR optics with a 26 dB power budget per link
(east or west). If necessary, optical power can be tuned with an optical
attenuator. The 15104, with its 3-R function, compensates for any dispersion
accumulated in the path. A similar concept applies to the STM-16 LR1
These are the specifications for the SMF LR1 interface:
Operating wavelength 1310 nm
Transmit power +2 dBm (max.) -3 dBm (min.)
Receive power -8 dBm (max.) -28 dBm (min.)
Recommended distance 40 km
Power budget 25 dB
Note: All DPT and RPR interfaces use SMF. Multi-Mode Fiber (MMF) is 850 nm
and with a core of 50 or 62.5 micron. The SMF is 1310 nm and 1550 nm with a
core of 8 micron.
How does DPT protection switching work?
A. The DPT/Resilient Packet Ring (RPR) protection switching uses a concept
similar to that of SONET or Synchronous Digital Hierarchy (SDH). The protection
switching is in a window of sub-50 msec switching. But, this does not use the
SONET or SDH detection parameters.
There are these three steps in case of a failure at a single ring
10 msec detection and sub-50 msec restoration (ring
Intelligent Protection Switching (IPS) topology update and
distribution for optimal path
Any route table update
The first two steps are very fast and belong to Layer 2 (L2) (SRP ,
Resource Availability Confirmation (RAC), Application-Specific Integrated
Circuit (ASIC), and the framer). The last step is in Layer 3 (L3) and is the
least to notice a topology change. Seldom does any single ring topology change,
due to a segment failure, trigger a route table update. This is because the
Layer 3 action is too slow, and most single rings use a single subnet. There is
no routing in such ring. There is never a race condition between SRP and any
Interior Gateway Protocol (IGP) or Exterior Gateway Protocol (EGP).
Multiprotocol Label Switching (MPLS) Fast Reroute (FRR) uses a similar
concept to that mentioned in step 1. If it is a very large network, such as a
long-haul DPT/RPR with dark fiber and cascaded 3-R regenerators, or as an
overlay over Dense Wavelength Division Multiplexing (DWDM), step 2 with the IPS
topology update and distribution for optimal path, takes extra time. There is
no interaction or any communication between any IGP or EGP, and SRP link
failure detection at the interface. The different layers are transparent and
such communication is for each particular layer end-to-end at each segment.
Typical restoration values are far less than 50 msec and are in the range of 5
to 10 msec in a lab environment (short spans). In the field this could be
different, but still less than 50 msec.
If there is transparency between the Layer 1 (L1), Layer 2, and Layer 3
failure detection mechanism, such as in the case of node, segment, or topology
failures, higher layers are not always aware. If Layer 1 handles the recovery
quickly, an Layer 2 mechanism such as Spanning Tree Protocol (STP), or an Layer
3 mechanism such as IGP or EGP does not do any restoration or reconverge. But,
some corner cases exist with DPT and RPR overlay and Packet over SONET (PoS)
What is DPT pass-through?
A. The interface can go in SRP pass-through under these two
If you put the interface in admin down
state with the shutdown command.
The MAC and Resource Availability Confirmation (RAC) watchdog
expires. The interface goes into the down state,
and the RAC and MAC is put in pass-through.
The srp shutdown [a|b] command is equivalent
to the srp ips request forced-switch [a|b] command,
and is not related to SRP pass-through mode.
This is a configuration sample:
Router-yb(config-if)#srp shutdown b
router-yb#show run int srp 1/1
no ip address
no ip directed-broadcast
srp ips request forced-switch b
Is Hot Standby Routing Protocol (HSRP) supported on Dynamic Packet
A. HSRP is not supported on SRP. The command line interface (CLI) command
you use in order to configure SRP has been disabled on the C10720, but this
does not look as though it was done on the Gigabit Switch Router (GSR). SRP
requires each node to have a single MAC address. But, with HSRP, you can assign
multiple MAC addresses to a single node which breaks this assumption. This can
work in certain setups, but this is not a stable configuration.