CCAMP Working Group CCAMP GMPLS P&R Design Team
Internet Draft
Expiration Date: December 2002 Dimitri Papadimitriou (Alcatel)
Eric Mannie (KPNQewst)
Deborah Brungard (AT&T)
Sudheer Dharanikota (Nayna)
Jonathan Lang (Calient)
Guangzhi Li (AT&T)
Bala Rajagopalan (Tellium)
Yakov Rekhter (Juniper)
June 2002
Analysis of GMPLS-based Recovery Mechanisms
(including Protection and Restoration)
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt
Status of this Memo
This document is an Internet-Draft and is in full conformance with
all provisions of Section 10 of RFC2026 [1].
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts. Internet-Drafts are draft documents valid for a maximum of
six months and may be updated, replaced, or obsoleted by other
documents at any time. It is inappropriate to use Internet- Drafts
as reference material or to cite them other than as "work in
progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
For potential updates to the above required-text see:
http://www.ietf.org/ietf/1id-guidelines.txt
1. Abstract
This document provides an analysis grid that can be used to
evaluate, compare and contrast the large amount of GMPLS based
recovery mechanisms currently proposed in the CCAMP WG. A detailed
analysis of each of the recovery phases as identified in [CCAMP-
TERM] will be given using terminology as defined in [CCAMP-TERM].
The focus will be on transport plane survivability and recovery
issues and ***not control plane resilience related aspects***.
D.Papadimitriou et al. - Internet Draft û Expires December 2002 1
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
2. Conventions used in this document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in
this document are to be interpreted as described in RFC-2119 [2].
3. Introduction
This document provides an analysis grid that can be used to
evaluate, compare and contrast the large amount of GMPLS based
recovery mechanisms currently proposed in the CCAMP WG. Here, the
focus will only be on transport plane survivability and recovery
issues and not on control plane resilience related aspects. Although
the recovery mechanisms described in this document impose different
requirements on recovery protocols, the protocol(s) specifications
will not be covered in this document. Despite the fact that the
concepts discussed here are technology independent, this document
will implicitly focus on SONET/SDH and pre-OTN technologies except
when specific details need to be considered (for instance, in the
case of failure detection). Details for applicability to other
technologies such as Optical Transport Networks (OTN) [ITUT-G709]
will be covered in a future release of this document.
In the present release, a detailed analysis is provided for each of
the recovery phases as identified in [CCAMP-TERM]. Recovery implies
that the following generic operations need to be performed when a
LSP/Span failure (or any other event generating such failures)
occurs:
- Phase 1: Failure detection
- Phase 2: Failure correlation
- Phase 3: Failure localization and isolation
- Phase 4: Failure notification
- Phase 5: Recovery (Protection/Restoration)
- Phase 6: Reversion (normalization)
Failure detection, correlation, localization and notification phases
together are referred to as fault management. Within a recovery
domain, the entities involved during the recovery operations are
defined in [CCAMP-TERM]; these entities include ingress, egress and
intermediate nodes.
In this document the term ôrecovery mechanismö will be used to cover
both protection and restoration mechanisms. The specific terms
protection and restoration will only be used when differentiation is
required. Likewise the term ôfailureö is used to represent both
signal failure and signal degradation. In addition, a clear
distinction will be made between partitioning (horizontal hierarchy)
and layering (vertical hierarchy). Any other recovery-related
terminology used in this document conforms to the one defined in
[CCAMP-TERM].
D.Papadimitriou et al. - Internet Draft û December 2002 2
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
4. Fault Management
4.1 Failure Detection
Transport failure detection is the only phase that can not be
achieved by the control plane alone since the latter needs a hook to
the transmission plane to gather the resulting information.
Therefore, by definition, failure detection is transport technology
dependent (and so exceptionally, we keep here the ôtransport planeö
terminology).
As an example, SONET/SDH (see [G.707], [G.783] and [G.806]) provides
supervision capabilities covering:
- Continuity: monitors the integrity of the continuity of a trail
(i.e. section or path). This operation is performed by monitoring
the presence/absence of the signal. Examples are Loss of Signal
(LOS) detection for the physical layer, Unequipped (UNEQ) Signal
detection for the path layer, Server Signal Fail Detection (e.g.
AIS) at the client layer.
- Connectivity: monitors the integrity of the routing of the signal
between end-points. Connectivity is normally only required if
the layer provides flexible connectivity, either automatically
(e.g. cross-connects controlled by the TMN) or manually (e.g.
fiber distribution frame). An example is the Trail (i.e. section,
path) Trace Identifier used at the different layers and the
corresponding Trail Trace Identifier Mismatch detection.
- Alignment: checks that the client and server layer frame start can
be correctly recovered from the detection of loss of alignment.
The specific processes depend on the signal/frame structure and
may include: (multi-)frame alignment, pointer processing and
alignment of several independent frames to a common frame start in
case of inverse multiplexing. Loss of alignment is a generic term.
Examples are loss of frame, loss of multi-frame, or loss of
pointer.
- Payload type: checks that compatible adaptation functions are used
at the source and the sink. This is normally done by adding a
signal type identifier at the source adaptation function and
comparing it with the expected identifier at the sink. For
instance, the payload signal label and the corresponding payload
signal mismatch detection.
- Signal Quality: monitors the performance of a signal. For
instance, if the performance falls below a certain threshold a
defect û excessive errors (EXC) or degraded signal (DEG) - is
detected.
The most important point to keep in mind is that the supervision
processes and the corresponding defect detection (used to initiate
the next recovery phase(s)) result in either:
D.Papadimitriou et al. - Internet Draft û December 2002 3
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
- Signal Degrade (SD): A signal indicating that the associated data
has degraded in the sense that a degraded defect condition is
active (for instance, a dDEG declared when the Bit Error Rate
exceeds a preset threshold).
- Signal Fail (SF): A signal indicating that the associated data has
failed in the sense that a signal interrupting near-end defect
condition is active (as opposed to the degraded defect).
In optical transport networks (OTN) equivalent supervision
capabilities are provided at the section layers (OTS, OMS and OTUk)
and at path layers (OCh and ODUk). Interested readers are referred
to the ITU-T Recommendations [G.798] and [G.709] for more details.
On the other hand, in pre-OTN networks, a failure may be masked by
O/E/O based OLS (Optical Line System), preventing Photonic Cross-
Connect (PXC) from detecting the failure. In such cases, failure
detection may be assisted by an out-of-band communication channel
and reported to the PXC control plane, such as considered in [LMP-
WDM]. The [LMP] protocol extensions it defines provides IP message-
based communication between the PXC and the OLS control plane. Also,
since PXCs are framing format independent, failure conditions can
only be triggered either by detecting the absence of the optical
signal or by measuring its quality. Both detection mechanisms are
out of the scope of this document. Using this communication channel,
these failure conditions are reported to the PXC and subsequent
recovery actions performed as described in Section 5. As such from
the control plane viewpoint, this mechanism makes the OLS-PXC
composed system appearing as a single logical entity.
More generally, the following are typical failure conditions in pre-
OTN networks:
- Loss of Light (LoL): signal failure condition where the optical
signal is not detected anymore on a given interfaceÆs receiver.
- Signal degradation (SD): detection of the signal degradation over
a specific period of time.
- For SDH/Sonet payloads, all of the above-mentioned supervision
capabilities can be used, resulting in SD or SF condition.
In summary, the following cases are considered to illustrate the
communication between detecting and reporting entities:
- Co-located detecting and reporting entities: both the detecting
and reporting entities are on the same node (e.g., SDH/SONET
equipment, Opaque cross-connects, and in some cases for
Transparent cross-connects, etc.).
- Non co-located detecting and reporting entities:
- with In-band communication between entities:
Entities are separated but in-band communication is provided
between them (e.g., APS, OXCÆs LOS, etc.).
- with Out-of-band communication between entities:
Entities are separated but out-of-band communication is provided
D.Papadimitriou et al. - Internet Draft û December 2002 4
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
between them (e.g., PXCÆs LOS, PXCÆs LOL, etc.).
4.2 Failure Correlation
A single failure (such as a span failure) can result into reporting
multiple failures (such as individual connection failures). Such
failures can be grouped i.e. correlated to reduce the communication
on the reporting channel, for both in-band and out-of-band failure
reporting.
In such a scenario, it can be important to wait for certain period
of time, typically called failure correlation time, and gather all
the failures to report them as a group of failures (or simply group
failure). For instance, this approach can be provided using LMP-WDM
for pre-OTN networks (see [LMP-WDM]) or when using Signal
Failure/Degrade Group in the SONET/SDH context.
Note that a default average time interval during which failure
correlation operation can be performed is difficult to provide since
strongly dependent on the underlying network topology. Therefore, it
can be advisable to provide a per node configurable failure
correlation time. The detailed selection criteria for this time
interval are outside of the scope of this document.
When failure correlation is not provided, multiple failure
indication messages may be sent out in response to a single failure
(for instance, a fiber), each one containing a set of information on
the failed working resources (for instance, the individual lambda
LSP flowing through this fiber). This allows for a more prompt
response but can potentially overload the control plane due to a
large amount of failure notifications.
4.3 Failure Localization and Isolation
The failure localization provides the information required in order
to perform the sub-sequent recovery action(s) at the LSP/span end-
points. However, in some cases, failure localization may be less
urgent. This is particularly the case when edge-to-edge LSP recovery
(edge referring to a domain end-node for instance) is performed
based on a simple failure notification (including the identification
of the failed working LSPs) while a more accurate localization can
be performed after subsequent LSP recovery.
Failure localization should be triggered immediately after fault
detection phase. This operation can be performed at the transport
management plane and/or the control plane level using dedicated
signaling messages.
When performed at the control plane level, a protocol such as LMP
(see [LMP], Section 6) can be used for failure localization and
isolation purposes.
4.4 Failure Notification
D.Papadimitriou et al. - Internet Draft û December 2002 5
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
Failure notification is used 1) to inform intermediate nodes that a
LSP/span failure has occurred and has been detected 2) to inform the
deciding entities (which can correspond to any intermediate or end-
point of the failed LSP/span) that the corresponding service is not
available. In general, these deciding entities will be the ones
taking the appropriate recovery decision. When co-located with the
recovering entity, they will also perform the corresponding recovery
action(s).
Failure notification can be either provided by the transport or by
the control plane. As an example, let us first briefly describe the
failure notification mechanism defined at the SDH/SONET transport
plane level (also referred to as maintenance signal supervision):
- AIS (Alarm Indication Signal) occurs as a result of a failure
condition such as Loss of Signal and is used to notify downstream
nodes (of the appropriate layer processing) that a failure has
occurred. AIS performs two functions 1) inform the intermediate
nodes (with the appropriate layer monitoring capability) that a
failure has been detected 2) notify the connection end-point that
the service is no longer available.
For a distributed control plane supporting one (or more) failure
notification mechanism(s), regardless of the mechanismÆs actual
implementation, the same capabilities are needed with more (or less)
information provided about the LSPs/Spans under failure condition,
their detailed status, etc.
The most important difference between these mechanisms is related to
the fact that transport plane notifications (as defined today) would
initiate a protection scheme (such as those defined in [CCAMP-TERM])
or a restoration scheme via the management plane. On the other hand,
using a failure notification mechanism through the control plane
would provide the possibility to trigger either a protection or a
restoration action via the control plane. Moreover, as specified in
[GMPLS-SIG], notification message exchanges through a GMPLS control
plane may not follow the same path as the LSP/spans they intent to
notify the unavailability. In turn, this ensures a reliable and
efficient failure notification mechanism.
The other important properties to be met by the failure notification
mechanism are mainly the following:
- Notification messages must provide enough information such that
the most efficient subsequent recovery action will be taken (in
most of the recovery schemes this action is even deterministic)
at the recovering entities. Remember here that the latter can be
either intermediate or end-points through which normal traffic
flows. Based on local policy, intermediate nodes may not use this
information for subsequent recovery actions (see for instance the
APS protocol phases as described in [CCAMP-TERM]).
D.Papadimitriou et al. - Internet Draft û December 2002 6
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
The trade-off here is to define what information the LSP/span end-
points (more precisely, the deciding entity) needs in order for
the recovering entity to take the best recovery action: if not
enough information is provided, the decision can not be optimal
(note that in this eventuality, the important issue is to quantify
the level of sub-optimality), if too much information is provided
the control plane may be overloaded with unnecessary information
and the aggregation/correlation of this notification information
will be more complex and time consuming to achieve. Notice that
more detailed quantification of the amount of information to be
exchanged and processed is strongly dependent on the failure
notification protocol specification.
- If the failure localization and isolation is not performed by one
of the LSP/Span end-points or some intermediate points, they
should receive enough information from the notification message in
order to locate the failure otherwise they would need to (re-)
initiate a failure localization and isolation action.
- Avoiding so-called notification storms implies that failure
detection output is correlated (i.e. alarm correlation) and
aggregated at the node detecting the failure(s), failure
notifications are directed to a restricted set of destinations (in
general the end-points), notification suppression (i.e. alarm
suppression) is provided in order to limit flooding in case of
multiple and/or correlated failures appearing at several locations
in the network
- Alarm correlation and aggregation (at the failure detecting
node) implies a consistent decisions based on the conditions for
which a trade-off between fast convergence (at detecting node) and
fast notification (implying that correlation and aggregation
occurs at receiving end-points) can be found.
5. Recovery Mechanisms and Schemes
5.1 Transport vs. Control Plane Responsibilities
TBD.
5.2 Technology in/dependent mechanisms
TBD.
5.3 Specific Aspects of Control Plane based Recovery Mechanisms
5.3.1 In-band vs Out-of-band Signalling
The nodes communicate through the use of IP control channels. Since
two classes of transport mechanisms can be considered here i.e. in-
band or out-of-band (through a dedicated physically diverse control
network), the potential impact of the signalling transport mechanism
is not a trivial issue.
D.Papadimitriou et al. - Internet Draft û December 2002 7
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
As such, the distinction between in-fiber in-band and in-fiber out-
of-band signalling reduces to the consideration of a logically
versus physically embedded control plane topology with respect to
the one of the transport plane. In the current scope of this
document, since we assume that IP control channels between nodes
must be continuously available in order to enable the exchange of
recovery-related information and messages, one considers that both
signalling transports provides at least either one logical channel
or one physical channel between nodes.
Therefore, the key issue when using in-band signalling is whether we
can assume independence between the fault-tolerance capabilities of
control plane and the failures affecting the transport plane
(including the nodes). Note also that existing specifications like
the OTN provide a limited form of independence for in-band signaling
by assigning control traffic to a separate supervisory optical
channel.
5.3.2 Uni- versus Bi-directional Failures
The failure detection, correlation and notification mechanisms
(described in Section 4) can be triggered when either a
unidirectional or a bi-directional LSP/Span failure occurs (or a
combination of both). As illustrated in Figure 1 and 2, two
alternatives can be considered here:
1. Uni-directional failure detection: the failure is detected on the
receiver side i.e. it is only is detected by the downstream node
to the failure (or by the upstream node depending on the failure
propagation direction, respectively)
2. Bi-directional failure detection: the failure is detected on the
receiver side of both downstream node AND upstream node to the
failure.
Notice that after the failure detection time, if only control plane
based failure management is provided, the peering node is unaware of
the failure detection status of its neighbor.
------- ------- ------- -------
| | | |Tx Rx| | | |
| NodeA |----...----| NodeB |xxxxxxxxx| NodeC |----...----| NodeD |
| |----...----| |---------| |----...----| |
------- ------- ------- -------
t0 >>>>>>> F
t1 x <---------------x
Notification
t2 <--------...--------x x--------...-------->
Up Notification Down Notification
D.Papadimitriou et al. - Internet Draft û December 2002 8
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
------- ------- ------- -------
| | | |Tx Rx| | | |
| NodeA |----...----| NodeB |xxxxxxxxx| NodeC |----...----| NodeD |
| |----...----| |xxxxxxxxx| |----...----| |
------- ------- ------- -------
t0 F <<<<<<< >>>>>>> F
t1 x <-------------> x
Notification
t2 <--------...--------x x--------...-------->
Up Notification Down Notification
Fig. 1 & 2. Uni- and Bi-directional Failure Detection/Notification
After failure detection, the following failure management operations
can be subsequently considered:
- Each detecting entity sends a notification message to the
corresponding transmitting entity. For instance, in Fig. 1 (Fig.
2), node C sends a notification message to node B (while node B
sends a notification message to node A). To ensure reliable
failure notification, a dedicated acknowledgment message can be
returned back to the sender node.
- Next, within a certain (and pre-determined) time window, nodes
impacted by the failure occurrences perform their correlation. In
case of unidirectional failure, node B only receives the
notification message from node C and thus the time for this
operation is negligible. However, in case of bi-directional
failure, node B (and node C) must correlate the received
notification message from node C (node B, respectively) with the
corresponding locally detected information.
- After some (pre-determined) period of time, referred to as the
hold-off time, after which local recovery actions were not
successful, the following occurs. In case of unidirectional
failure and depending on the directionality of the connection,
node B should send an upstream notification message to the ingress
node A or node C should send a downstream notification to the
egress node D. However, in such a case only node A (node D,
respectively) would initiate a edge to edge recovery action. Note
that the connection terminating node (i.e. node D or node A) may
be optionally notified.
In case of bi-directional failure, and depending on the
directionality of the connection, node B may send an upstream
notification message to the ingress node A or node C a downstream
notification to the egress node D. However, due to the dependence
on the connection directionality, only ingress node A or egress
D.Papadimitriou et al. - Internet Draft û December 2002 9
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
node D would initiate an edge to edge recovery action. Note that
the connection terminating node (i.e. node D or node A) should be
notified. For instance, if a connection directed from D to A is
under failure condition, only the notification sent by from node C
to D would initiate a recovery action.
In the above scenarios, the path followed by the notification
messages does not have to be the same as the one followed by the
failed LSP (see [GMPLS-SIG], for more details on the notification
message exchange). The important point, concerning this mechanism,
is that either the detecting/reporting entity (i.e. the nodes B and
C) are also the deciding/recovery entity or the detecting/reporting
entities are simply intermediate nodes in the subsequent recovery
process. One refers to local recovery in the former case and to
edge-to-edge recovery in the latter one.
5.3.3 Partial versus Full Span Recovery
When given span carries more than one LSP or LSP segment, an
additional aspect must be considered during span failure carrying
several LSPs. These LSP can be either individually recovered or
recovered as a group (bulk LSP recovery) or independent sub-groups.
The selection of this mechanism would be triggered independently of
the failure notification granularity when correlation time windows
are used and simultaneous recovery of several LSP can be performed
using single request. Moreover, criteria by which such sub-groups
can be formed are outside of the scope of this document.
An additional complexity arises in case of recovery a LSP group. The
LSPs created between a node pair may have been initiated from
different source (i.e. initiator) nodes. Consequently the node
downstream to a bi-directional span failure affecting several LSPs
(or the whole group of LSP it carries) is not necessarily directed
toward the same destination node. Therefore, such span failure may
generate recovery actions to be performed a both LSP initiator nodes
of the pair. In order to facilitate the definition of the recovery
mechanisms (and their sequence) one assumes here that the initiator
of the LSP/LSP Segment is also the deciding entity (see [CCAMP-
TERM]) for its recovery.
5.3.4 Difference between LSP, LSP Segment and Span Recovery
The recovery definitions given in [CCAMP-TERM] are quite generic and
apply for link (or local span) and LSP recovery. The major
difference between LSP, LSP Segment and span recovery is related to
the number of intermediate nodes the signalling messages have to
travel. Since nodes are not necessarily adjacent in case of LSP (or
LSP Segment) recovery, signalling message exchanges from the
reporting to the deciding/recovery entity will have to cross several
intermediate nodes. In particular, this applies for the notification
messages due to the number of hops separating the failure occurrence
location from their destination. This results in an additional
propagation and forwarding delay, which can in certain circumstances
D.Papadimitriou et al. - Internet Draft û December 2002 10
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
be non-negligible e.g. in case of copper out-of-band network one has
to consider approximately 1 ms per 200km.
Moreover, the recovery mechanisms applicable to end-to-end LSP and
to the segments (i.e. edge-to-edge) that may compose an end-to-end
LSP can be exactly the same. However, one expects in the latter
case, that the destination of the failure notification message will
be the ingress of each of these segments. Therefore, taking into
account the mechanism described in Section 5.3.2, failure
notification can be first exchanged between LSP segment terminating
points and after expiration of the hold-off time directed toward
end-to-end LSP terminating points.
5.4 Difference between Recovery Type and Scheme
Section 4.6 of [CCAMP-TERM] defines the basic recovery types. The
purpose of this section is to describe the schemes that can be built
using these recovery types. Several examples are provided in order
to illustrate the difference between a recovery type such as 1:1 and
a recovery scheme such as (1:1)^n.
1. (1:1)^n with recovery resource sharing
The exponent, n, indicates the number of times a 1:1 recovery type
is applied between at most n different ingress-egress node pairs.
Here, at most n pairs of disjoint recovery and working LSPs/spans
share at most n times a unique common resource. Since the working
LSPs/Spans are mutually disjoint, simultaneous requests for use of
the shared resource will only occur in case of a simultaneous
failures, which is less likely to happen.
For instance, in the (1:1)^2 common case if the 2 recovery LSPs in
the group overlap the same common resource, then it can handle only
single failures; any multiple working LSP failures will cause at
least one working LSP to be denied automatic recovery. Consider for
instance, the following example, with working LSPs A-B and E-F and
recovery LSPs A-C-D-B and E-C-D-F sharing a common C-D resource.
A --------------- B
\ /
C ----------- D
/ \
E --------------- F
2. (M:N)^n with recovery resource sharing
The exponent, n, indicates the number of times a M:N recovery type
is applied between at most n different ingress-egress node pairs.
So the interpretation follows from the previous case, expect that
here disjointness applies to the N working LSPs/spans and to the M
recovery LSPs/spans while sharing at most (n x M) common resources.
D.Papadimitriou et al. - Internet Draft û December 2002 11
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
In both schemes, one may see the following at the LSP level: we have
a ôgroupö of sum{n=1}^N N{n} working LSPs and a pool of shared
backup resources, not all of which are available to any given
working path. In such conditions, defining a metric that describes
the amount of overlap among the recovery LSPs would give some
indication of the groupÆs ability to handle multiple simultaneous
failures.
For instance, in the simpler (1:1)^n case situation if n recovery
LSPs in a (1:1)^n group overlap, then it can handle only single
failures; any multiple working LSP failures will cause at least one
working LSP to be denied automatic recovery. But if one consider for
instance, a (1:1)^4 group in which there are two pairs of
overlapping recovery LSPs, then two LSPs (belonging to the same
pair) can be simultaneously recovered. The latter case can be
illustrated as follows: 2 working LSPs A-B and E-F and 2 recovery
LSPs A-C-D-B and E-C-D-F sharing the two common C-D resources.
A ================ B
\\ //
C =========== D
// \\
E ================ F
Moreover, in all these schemes, (working) path disjointness can be
reinforced by exchanging working LSP related information during the
recovery LSP signalling.
5.5 LSP Restoration Schemes
5.5.1 Classification
LSPs/spans recovery time depends on the proper recovery LSP (soft)
provisioning and the level of recovery resources overbooking (i.e.
over-provisioning). A proper balance of these two mechanisms will
result in a desired LSP/span recovery time when single or multiple
failure(s) occur(s).
Recovery LSP Provisioning phases:
(1) Route Computation --> On-demand
|
|
--> Pre-Computed
|
|
(2) Signalling --> On-demand
|
|
--> Pre-Signaled
|
|
(3) Resource Selection --> On-demand
D.Papadimitriou et al. - Internet Draft û December 2002 12
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
|
|
--> Pre-Selected
Overbooking Levels:
+----- Dedicated (for instance: 1+1, 1:1, etc.)
|
|
+----- Shared (for instance: 1:N, M:N, etc.)
|
Level of |
Overbooking -----+----- Unprotected (for instance: 0:1, 0:N)
Fig 3. LSP Provisioning and Overbooking Classification
In this figure, we present a classification of different options
under LSP provisioning and overbooking. Although we acknowledge
these operations are run mostly during planning (using network
planning) and provisioning time (using signaling and routing)
activities, we keep them in analyzing the recovery schemes.
Proper LSP/span provisioning will help in alleviating many of the
failures. As an example, one may compute primary and secondary
paths, either end-to-end or segment-per-segment, to recover an LSP
from multiple failure events affecting link(s), node(s), SRLG(s)
and/or SRG(s). Such primary and secondary LSP/span provisioning can
be categorized, as shown in the above figure, based on:
(1) the recovery path (i.e. route) can be either pre-computed or
computed on demand.
(2) when the recovery path is pre-computed: pre-signaled (implying
recovery resource reservation) or signaled on demand.
(3) and when the recovery resources are reserved, they can be either
pre-selected or selection on-demand.
Note that these different options give rise to different LSP/span
recovery times. The following subsections will consider all these
cases in analyzing the schemes.
There are many mechanisms available allowing the overbooking of the
recovery resources. This overbooking can be done per LSP (such as
the example mentioned above), per link (such as span protection) or
per domain (such as ring topologies). In all these cases the level
of overbooking, as shown in the above figure, can be classified as
dedicated (such as 1+1 and 1:1), shared (such as 1:N and M:N) or
unprotected (i.e. restorable if enough recovery resources are
available).
Under a shared restoration scheme one may support preemptable
(preempt low priority connections in case of resource contention)
D.Papadimitriou et al. - Internet Draft û December 2002 13
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
extra-traffic. In this document we keep in mind all the above-
mentioned overbooking mechanisms in analyzing the recovery schemes.
5.5.2 Dynamic LSP Restoration
We first define the following times in order to provide a
quantitative estimation about the time performance of the dynamic
and pre-signaled LSP restoration:
- Path Computation Time - Tpc
- Path Selection Time - Tps
- End-to-end LSP Resource Reservation û Trr (a delta for resource
selection is also considered, the total time is then referred to
as Trs)
- End-to-end LSP (Resource) Activation Time û Tra (a delta for
resource selection is also considered, the total time is then
referred to as Tas)
Note: failure management operations such as failure detection,
correlation and notification are considered as equivalently time
consuming for all the mechanisms described here below:
1. With Route Pre-computation
An end-to-end restoration LSP is established after the failure(s)
occur(s) based on a pre-computed path (i.e. route). As such, one can
define this as an ôLSP re-provisioningö mechanism. Here, one or more
(disjoint) routes for the restoration path are computed (and
optionally pre-selected) before a failure occurs.
No reservation or selection of resources is performed along the
restoration path before failure. As a result, there is no guarantee
that a restoration connection is available when a failure occurs.
The total time T expected is thus equal to Tps + Trs or when a
dedicated computation is performed for each working LSP to Trs.
2. Without Route Pre-computation
An end-to-end restoration LSP is established after the failure(s)
occur(s). Here, one or more (disjoint) explicit route for the
restoration path are dynamically computed and one is selected after
failure. As such, one can define this as an ôLSP re-provisioningö
mechanism.
No reservation or selection of resources is performed along the
restoration path before failure. As a result, there is no guarantee
that a restoration connection is available when a failure occurs.
The total time T expected is thus equal to Tpc + Tps + Trs.
Therefore, time performance between the two approaches differs only
by the time required for route computation (and selection).
5.5.3 Pre-signaled Restoration LSP
D.Papadimitriou et al. - Internet Draft û December 2002 14
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
1. With resource reservation without pre-selection
An end-to-end restoration path is pre-selected from a set of one or
more pre-computed (disjoint) explicit route before failure. The
restoration LSP is signaled along this pre-selected path to reserve
resources (i.e. signaled) at each node but resources are not
selected.
In this case, the resources reserved for each restoration LSP may be
dedicated or shared between different working LSP that are not
expected to fail simultaneously. Local node policies can be applied
to define the degree to which these resources are shared across
independent failures.
Upon failure detection, signaling is initiated along the restoration
path to select the resources, and to perform the appropriate
operation at each node entity involved in the restoration connection
(e.g. cross-connections).
The total time T expected is thus equal to (Tps +) Trr + Tas.
2. With resource reservation and pre-selection
An end-to-end restoration path is pre-selected from a set of one or
more pre-computed (disjoint) explicit route before failure. The
restoration LSP is signaled along this pre-selected path to reserve
AND select resources at each node but not cross-connected. Such that
the selection of the recovery resources is fixed at the control
plane level. However, no cross-connections are performed along the
restoration path.
In this case, the resources reserved for each restoration LSP may
only be shared between different working LSPs that are not expected
to fail simultaneously. Since one considers restoration schemes
here, the sharing degree should not be limited to working (and
recovery) LSPs starting and ending at the same ingress and egress
nodes. Therefore, one expects to receive some feedback information
on the recovery resource sharing degree at each node participating
to the recovery scheme.
Upon failure detection, signaling is initiated along the restoration
path to activate the reserved and selected resources and to perform
the appropriate operation at each node involved in the restoration
connection (e.g. cross-connections).
The total time T expected is thus equal to (Tps +) Trs + Tra.
Therefore, time performance between the two approaches differs only
by the time required for resource selection during the activation of
the recovery LSP.
5.5.4 LSP Segment Restoration
D.Papadimitriou et al. - Internet Draft û December 2002 15
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
The above approaches can be applied on a sub-network basis rather
than end-to-end basis (in order to reduce the global recovery time).
It should be also noted that using the horizontal hierarchical
approach described in Section 7.1, that a given end-to-end LSP can
be recovered by multiple recovery mechanisms (e.g. 1:1 protection in
a metro edge network but M:N protection in the core). These
mechanisms are ideally independent and may even use different
failure localization and notification mechanisms.
6. Normalization
6.1 Wait-To-Restore
A specific mechanism (Wait-To-Restore) is used to prevent frequent
protection switching operation due to an intermittent defect (e.g.
BER fluctuating around the SD threshold).
First, a failed LSP/span must become fault-free, e.g. a BER less
than a certain recovery threshold. After the recovered LSP/span
(i.e. the previously working LSP/span) meets this criterion, a fixed
period of time shall elapse before a normal traffic uses the
corresponding resources again. This period called Wait-To-Restore
(WTR) period or timer is generally of the order of a few minutes
(for instance, 5 minutes) and should be capable of being set. An SF
or SD condition overrides the WTR.
6.2 Revertive Mode Operation
In revertive mode of operation, when the recovery LSP/span is no
longer required, i.e. the failed working LSP/span is no longer in SD
or SF condition, a local Wait-to-Restore (WTR) state will be
activated before switching the normal traffic back to the recovered
working LSP/span.
During the reversion operation, since this state becomes the highest
in priority, signalling must maintain the normal traffic on the
recovery LSP/span from the previously failed working LSP/span.
Moreover, during this WTR state, any null traffic or extra traffic
(if applicable) request is rejected. However, deactivation of the
wait-to-restore timer may occur in case of higher priority request
attempts.
6.3 Orphans
When a reversion operation is requested normal traffic must be
switched from the recovery to the ôrecoveredö working LSP/span. A
particular situation occurs when the working LSP/span can not be
recovered such that normal traffic can not be switched back. In such
a case, the unrecoverable working LSP/span or segment (also referred
to as ôorphanö) must be cleared. Otherwise, potential de-
synchronization between the control and transport plane resource
D.Papadimitriou et al. - Internet Draft û December 2002 16
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
usage can appear. Depending on the signalling protocol capabilities
and behavior different mechanisms are to be expected here.
Several ways can be used for that purpose: wait for the elapsing of
the clear-out time interval, initiate a deletion from the ingress or
the egress node or trigger the initiation of deletion from an entity
(such as an EMS or NMS) capable to react on the reception of an
appropriate notification message.
7. Hierarchies
Recovery mechanisms are being made available at multiple (if not
each) transport layers within so-called ôIP-over-opticalö networks.
However, each layer has certain recovery features and one needs to
determine the exact impact of the interaction between the recovery
mechanisms provided by these layers.
Hierarchies are used to build scalable complex systems. Abstraction
is used as a mechanism to build large networks or as a technique for
enforcing technology, topological or administrative boundaries. The
same hierarchical concept can be applied to control the network
survivability. In general, it is expected that the recovery action
is taken by the recoverable LSP/span closest to the failure in order
to avoid the multiplication of recovery actions. Moreover, recovery
hierarchies can be also bound to control plane logical partitions
(e.g. administrative or topological boundaries). Each of them may
apply different recovery mechanisms.
In brief, commonly accepted ideas are generally that the lower
layers can provide coarse but faster recovery while the higher
layers can provide finer but slower recovery. Moreover, it is also
more than desirable to avoid too many layers with functional
overlaps. In this context, this section intends to analyze these
hierarchical aspects including the physical (passive) layer(s).
7.1 Horizontal Hierarchy (Partitioning)
A horizontal hierarchy is defined when partitioning a single layer
network (and its control plane) into several recovery domains.
Within a domain, the recovery scope may extend over a link (or
span), LSP segment or even an end-to-end LSP. Moreover, an
administrative domain may consist of a single recovery domain or can
be partitioned into several smaller recovery domains. The operator
can partition the network into recovery domains based on physical
network topology, control plane capabilities, or various traffic
engineering constraints.
An example often addressed in the literature is the metro-core-metro
application (sometimes extended to a metro-metro/core-core) within a
single transport layer (see Section 7.2). For such a case, an end-
to-end LSP is defined between the ingress and egress metro nodes,
while LSP segments may be defined within the metro or core sub-
networks. Each of these topological structures determines a so-
D.Papadimitriou et al. - Internet Draft û December 2002 17
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
called ôrecovery domainö since each of the LSPs they carry can have
its own recovery type (or even scheme). The support of multiple
recovery schemes within a sub-network is referred to as a multi-
recovery capable domain or simply multi-recovery domain.
7.2 Vertical Hierarchy (Layers)
It is a very challenging task to combine in a coordinated manner the
different recovery capabilities available across the path (i.e.
switching capable) and section layers to ensure that certain network
survivability objectives are met for the different services
supported by the network.
As a first analysis step, one can draw the following guidelines for
a vertical coordination of the recovery mechanisms:
- The lower the layer the faster the notification and switching
- The higher the layer the finer the granularity of the recoverable
entity and therefore the granularity of the recovery resource
(and subsequently its sharing ratio)
A vertical hierarchy consists of multiple layered transport planes
providing different:
- Discrete bandwidth granularities for non-packet LSPs such as OCh,
ODUk, HOVC/STS-SPE and LOVC/VT-SPE LSPs and continuous bandwidth
granularities for packet LSPs
- Potentially, recovery capabilities with different temporal
granularities: ranging from milliseconds to tens of seconds
In SDH/Sonet environments, one typically considers the LOVC/VT and
HOVC/STS SPE as independent layers, LOVC/VT LSP using the underlying
HOVC/STS SPE LSPs as links, for instance. In OTN, the ODUk path
layers will lie on the OCh path layer i.e. the ODUk LSPs using the
underlying OCh LSPs as links. Notice here that server layer LSPs may
simply be provisioned and not dynamically triggered or established
(control driven approach).
The following figure (including only the path layers) illustrates
the hierarchical layers that can be covered by the recovery
architecture of a transmission network comprising a SDH/Sonet and an
OTN part:
LOVC <------------------------------------------------------> LOVC
|| ||
HOVC ==== HOVC <----------------------------------> HOVC ==== HOVC
|| ||
ODUk ==== ODUk <--------------> ODUk ==== ODUk
|| ||
OCh <---> OCh <---> OCh
In this context, the important points are the following:
- these layers are path layers; i.e. the ones controlled by
the GMPLS (in particular, signalling) protocol suite.
- an LSP at the lower layer for instance an optical channel (=
D.Papadimitriou et al. - Internet Draft û December 2002 18
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
network connection) appears as a section (= link) for the OTUk
layer i.e. the links that are typically controlled by link
management protocols such as LMP.
If one considers also the section layers of the OTH then the
following scheme applies:
ODUk == . . == ODUk <-------------------------> ODUk == . . == ODUk
|| ||
OTUk <-------------------------> OTUk
|| ||
OCh <---> OCh <-...-> OCh <---> OCh
The first key issue with multi-layer recovery is that achieving
control plane individual or bulk LSP recovery will be as efficient
as the underlying link (local span) recovery. In such a case, the
span can be either protected or unprotected, but the LSP it carries
MUST be (at least locally) recoverable. Therefore, the span recovery
process can either be independent when protected (or restorable), or
triggered by the upper LSP recovery process. The former requires
coordination in order to achieve subsequent LSP recovery. Therefore,
in order to achieve robustness and fast convergence, multi-layer
recovery requires a fine-tuned coordination mechanism.
Moreover, in the absence of adequate recovery mechanism coordination
(pre-determined for instance by the hold-off time), a failure
notification may propagate from one layer to the next within a
recovery hierarchy. This can cause "collisions" and trigger
simultaneous recovery actions that may lead to race conditions and
in turn, reduce the optimization of the resource utilization and/or
generate global instabilities in the network (see [MANCHESTER]).
Therefore, a consistent and efficient escalation strategy is needed
to coordinate recovery across several layers.
Therefore, one can expect that the definition of the recovery
mechanisms and protocol(s) is technology independent such that they
can be implemented at different layers; this would in turn simplify
their global coordination.
Note: Recovery Granularity
In most environments, the design of the network and the vertical
distribution of the LSP bandwidth are such that the recovery
granularity is finer for higher layers. The OTN and SDH/Sonet layers
can only recover the whole section or the individual connections it
transports whereas IP/MPLS layer(s) can recover individual packet
LSPs or groups of packet LSPs.
Obviously, the recovery granularity at the sub-wavelength (i.e.
SDH/Sonet) level can be provided only when the network includes
devices switching at the same granularity level (and thus not with
optical channel switching capable devices). Therefore, the network
layer can deliver control-plane driven recovery mechanisms on a per-
D.Papadimitriou et al. - Internet Draft û December 2002 19
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
LSP basis if and only if the LSPs class has the corresponding
switching capability at the transport plane level.
7.3 Escalation Strategies
There are two types of escalation strategies (see [DEMEESTER]):
bottom-up and top-down.
The bottom-up approach assumes that lower layer recovery schemes are
more expedient and faster than the upper layer one. Therefore we can
inhibit or hold-off higher layer recovery. However this assumption
is not entirely true. Imagine a SDH/Sonet based protection mechanism
(with a less than 50 ms protection switching time) lying on top of
an OTN restoration mechanism (with a less than 200 ms restoration
time). Therefore, this assumption should be (at least) clarified as:
lower layer recovery schemes are faster than upper level one but
only if the same type of recovery mechanism is used at each layer
(assuming that the lower layer one is faster).
Consequently, taking into account the recovery actions at the
different layers in a bottom-up approach, if lower layer recovery
mechanisms are provided and sequentially activated in conjunction
with higher layer ones, the lower layers MUST have an opportunity to
recover normal traffic before the higher layers do. However, if
lower layer recovery is slower than higher layer recovery, the lower
layer MUST either communicate the failure related information to the
higher layer(s) (and allow it to perform recovery), or use a hold-
off timer in order to temporarily set the higher layer recovery
action in a ôstandby modeö. Note that the a priori information
exchange between layers concerning their efficiency is not within
the current of this document. Nevertheless, the coordination
functionality between layers must be configurable and tunable.
An example of coordination between the optical and packet layer
control plane enables for instance letting the optical layer
performing the failure management operations (in particular, failure
detection and notification) while giving to the packet layer control
plane the authority to perform the recovery actions. In case of
packet layer unsuccessful recovery action, fallback at the optical
layer can be subsequently performed.
The Top-down approach attempts service recovery at the higher layers
before invoking lower layer recovery. Higher layer recovery is
service selective, and permits "per-CoS" or "per-connection" re-
routing. With this approach, the most important aspect is that the
upper layer must provide its own reliable and INDEPENDENT failure
detection mechanism from the lower layer.
The same reference suggests also recovery mechanisms incorporating a
coordinated effort shared by two adjacent layers with periodic
status updates. Moreover, at certain layers, some of these recovery
operations can be pre-assigned, e.g. a particular link will be
D.Papadimitriou et al. - Internet Draft û December 2002 20
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
handled by the packet layer while another will be handled by the
fiber layer.
7.4 Disjointness
Having link and node diverse working and recovery LSPs/spans does
not guarantee working and recovery LSPs/Spans disjointness. Due to
the common physical layer topology (passive), additional
hierarchical concepts such as the Shared Risk Link Group (SRLG) and
mechanisms such as SRLG diverse path computation must be developed
to provide a complete working and recovery LSP/span disjointness
(see [IPO-IMP] and [CCAMP-SRLG]). Otherwise, a failure affecting the
working LSP/span would also potentially affect the recovery LSP/span
resources, one refers to such event as a common failure.
7.4.1 SRLG Disjointness
A Shared Risk Link Group (SRLG) is defined as the set of optical
spans (or links or optical lines) sharing a common physical resource
(for instance, fiber links, fiber trunks or cables) i.e. sharing a
common risk. For instance, a set of links L belongs to the same SRLG
s, if they are provisioned over the same fiber link f.
The SRLG properties can be summarized as follows:
1) A link belongs to more than one SRLG if and only if it crosses
one of the resources covered by each of them.
2) Two links belonging to the same SRLG can belong individually to
other (one or more) SRLGs.
3) The SRLG set S of an LSP is defined as the union of the
individual SRLG s of the individual links composing this LSP.
SRLG disjointness for LSP:
The LSP SRLG disjointness concept is based on the following
postulate: an LSP (i.e. sequence of links) covers an SRLG if and
only if it crosses one of the links belonging to that SRLG.
Therefore, the SRLG disjointness for LSPs can be defined as
follows: two LSPs are disjoint with respect to an SRLG s if and
only if none of them covers simultaneously this SRLG.
While the LSP SRLG disjointness with respect of a set S of SRLGs
is defined as follows: two LSPs are disjoint with respect to a
set of SRLGs S if and only if the sets of SRLGs they cover are
completely and mutually disjoint.
The impact on recovery is obvious: SRLG disjointness is a necessary
(but not a sufficient) condition to ensure optical network
survivability. With respect to the physical network resources, a
working-recovery LSP/span pair must be SRLG disjoint in case of
D.Papadimitriou et al. - Internet Draft û December 2002 21
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
dedicated recovery type while a working-recovery LSP/span group must
be SRLG disjoint in case of shared recovery.
7.4.2 SRG Disjointness
By extending the previous definition from a link to a more generic
structure, referred to as a ôrisk domainö, one comes to the SRG
(Shared Risk Group) notion (see [CCAMP-SRG]). A risk domain is a
group of arbitrarily connected nodes and spans that together can
provide certain like-capabilities (such as a chain of dedicated/
shared protected links and nodes, or a ring forming nodes and links,
or a protected hierarchical TE Link).
In turn, an SRG represents the risk domain capabilities and other
parameters, which assist in computing diverse paths through the
domain (it can also be used in assessing the risk associated with
the risk domain.)
Note that the SRLG set of a risk domain constitutes a subset of the
SRGs. SRLGs address only risks associated with the links (physical)
and passive elements within the risk domain, whereas SRGs may
contain nodes and other topological information in addition to the
links. The key difference between an SRLG and an SRG is that an SRLG
translates to only one link share risk with respect to server layer
topology (even hierarchical TE Links) while an SRG translates a
sequence of SRLGs over the same layer from one source to one or more
than one destination located within the same area.
As for SRLG disjointness, the impact on recovery is that SRG
disjointness is a necessary (but not a sufficient) condition to
ensure optical network survivability. With respect to the physical
and logical network resources (and topology), a working-recovery
LSP/span pair must be SRG disjoint in case of dedicated recovery
type while a working-recovery LSP/span group must be SRG disjoint in
case of shared recovery.
8. Recovery Scheme/Strategy Selection
In order to provide a structured selection and analysis of the
recovery scheme/strategy, the following dimensions can be defined:
1. Fast convergence (performance): provide a mechanism that
aggregates multiple failures (this implies fast failure
detection and correlation mechanisms) and fast recovery decision
independently of the number of failures occurring in the optical
network (implying also a fast failure notification).
2. Efficiency (scalability): minimize the switching time required
for LSP/span recovery independently of number of LSPs/spans being
recovered (this implies an efficient failure correlation, a fast
failure notification and timely efficient recovery mechanism(s)).
3. Robustness (availability): minimize the LSP/span downtime
D.Papadimitriou et al. - Internet Draft û December 2002 22
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
independently of the underlying topology of the transport plane
(this implies a highly responsive recovery mechanism).
4. Resource optimization (optimality): minimize the resource
capacity, including LSP/span and nodes (switching capacity),
required for recovery purposes; this dimension can also be
referred to as optimize the sharing degree of the recovery
resources.
5. Cost optimization: provide a cost-effective recovery strategy.
However, these dimensions are either out of the scope of this
document such as cost optimization, recovery path computational
aspects or going in opposite directions. For instance, it is obvious
that providing a 1+1 recovery type for each LSP minimizes the LSP
downtime (in case of failure) while being non-scalable and recovery
resource consuming without enabling any extra-traffic.
The following sections try to provide a first response in order to
select a recovery strategy with respect to the dimensions described
above and the recovery schemes proposed in [CCAMP-TERM].
8.1 Fast Convergence (Detection/Correlation and Hold-off Time)
Fast convergence is related to the failure management operations. It
refers to the elapsing time between the failure detection/
correlation and hold-off time, point at which the recovery switching
actions are initiated. This point has been already discussed in
Section 4.
8.2 Efficiency (Switching Time)
In general, the more pre-assignment/pre-planning of the recovery
LSP/span, the more rapid the recovery scheme is. Since protection
implies pre-assignment (and cross-connection in case of LSP
recovery) of the protection resources, in general, protection
schemes recover faster than restoration schemes.
Span restoration (since using control plane) is also likely to be
slower than most span protection types; however this greatly depends
on the span restoration signalling efficiency. LSP Restoration with
pre-signaled and pre-selected recovery resources is likely to be
faster than fully dynamic LSP restoration, especially because of the
elimination of any potential crank-back during the recovery LSP
establishment.
If one excludes the crank-back issue, the difference between dynamic
and pre-planned restoration depends on the restoration path
computation and path selection time. Since computational
considerations are outside of the scope of this document, it is up
to the vendor to determine the average path computation time in
different scenarios and to the operator to decide whether or not
dynamic restoration is advantageous over pre-planned schemes
D.Papadimitriou et al. - Internet Draft û December 2002 23
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
depending on the network environment. This difference depends also
on the flexibility provided by pre-planned restoration with respect
to dynamic one: the former implies a limited number of failure
scenarios (that can be due for instance to local storage
limitation). This, while the latter enables an on-demand path
computation based on the information received through failure
notification and as such more robust with respect to the failure
scenario scope.
Moreover, LSP segment restoration, in particular, dynamic
restoration (i.e. no path pre-computation so none of the recovery
resource is pre-signaled) will generally be faster than end-to-end
LSP schemes. However, local LSP restoration assumes that each LSP
segment end-point has enough computational capacity to perform this
operation while end-to-end requires only that LSP end-points
provides this path computation capability.
Recovery time objectives for SDH/Sonet protection switching (not
including time to detect failure) are specified in [G.841] at 50 ms,
taking into account constraints on distance, number of connections
involved, and in the case of ring enhanced protection, number of
nodes in the ring. Recovery time objectives for restoration
mechanisms have been proposed through a separate effort [TE-RH].
8.3 Robustness
In general, the less pre-assignment (protection)/pre-planning
(restoration) of the recovery LSP/span, the more robust the recovery
type/scheme is to a variety of (single) failures, provided that
adequate resources are available. Moreover, the pre-selection of the
recovery resources gives less flexibility for multiple failure
scenarios than no recovery resource pre-selection. For instance, if
failures occur that affect two LSPs sharing a common link along
their restoration paths, then only one of these LSPs can be
recovered. This occurs unless the restoration path of at least one
of these LSPs is re-computed or the local resource assignment is
modified on the fly.
In addition, recovery schemes with pre-planned recovery resources,
in particular spans for protection and LSP for restoration purposes,
will not be able to recover from failures that simultaneously affect
both the working and recovery LSP/span. Thus, the recovery resources
should ideally be chosen to be as disjoint as possible (with respect
to link, node and SRLG) from the working ones, so that any single
failure event will not affect both working and recovery LSP/span. In
brief, working and recovery resource must be fully diverse in order
to guarantee that a given failure will not affect simultaneously the
working and the recovery LSP/span. Also, the risk of simultaneous
failure of the working and restoration LSP can be reduced by re-
computing a restoration path whenever a failure occurs along the
corresponding recovery LSP or by re-computing a restoration path and
re-provisioning the corresponding recovery LSP whenever a failure
D.Papadimitriou et al. - Internet Draft û December 2002 24
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
occurs along a working LSP/span. This method enables to maintain the
number of available recovery path constant.
The robustness of a recovery scheme is also determined by the amount
of reserved (i.e. signaled) recovery resources within a given shared
resource pool: as the amount of recovery resources sharing degree
increases, the recovery scheme becomes less robust to multiple
failure occurrences. Recovery schemes, in particular restoration,
with pre-signaled resource reservation (with or without pre-
selection) should be capable to reserve the adequate amount of
resource to ensure recovery from any specific set of failure events,
such as any single SRLG failure, any two SRLG failures etc.
8.4 Resource Optimization
It is commonly admitted that sharing recovery resources provides
network resource optimization. Therefore, from a resource
utilization perspective, protection schemes are often classified
with respect to their degree of sharing protection resources with
respect to the working entities. Moreover, non-permanent bridging
protection types allow (under normal conditions) for extra-traffic
over the recovery resources. 1+1 LSP/Span protection is the more
resource consuming protection type since it doesnÆt allow for any
extra-traffic. 1:1 and 1:N LSP/span protection types require
dedicated recovery LSP/span while allowing extra (preemptible)
traffic, shared between the N working LSP/span in case of 1:N
protection. Obviously, 1+1 and 1:1 protection types do not provide
protection resource sharing while 1:N and M:N protection type allow
sharing of 1 (M, respectively) protection LSP/spans between N
working LSP/spans. However the flexibility in usage of shared
protection resources (in particular, shared protection links) may be
limited because of network topology restrictions, e.g. fixed ring
topology for traditional enhanced protection schemes.
On the other hand, the degree to which restoration schemes allow
sharing amongst multiple independent failures is directly dictated
by the size of the restoration pool. In restoration schemes with re-
provisioning, a pool of restoration resource can be defined from
which all restoration routes are selected after failure occurrence.
Thus, the degree of sharing is defined by the amount of available
restoration capacity. In restoration with pre-signaled resource
reservation, the amount of reserved restoration capacity is
determined by the local bandwidth reservation policies. In all
restoration schemes, preemptible LSP/span can use spare restoration
resources when these resources are not being used for LSP/span
recovery purposes.
Clearly, less recovery resources (i.e. LSP/spans and switching
capacity) have to be allocated to a shared recovery source pool if a
greater sharing degree is required. Thus, the degree to which the
network is survivable is determined by the policy that defines the
amount of reserved (shared) recovery resources.
D.Papadimitriou et al. - Internet Draft û December 2002 25
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
8.4.1. Recovery Resource Sharing
When recovery resources are shared over several LSP/Spans, [GMPLS-
RTG] through the use of the Maximum LSP Bandwidth, the Maximum
Reservable Bandwidth and the Unreserved Bandwidth TE Link sub-TLVs
provides the required parameters to obtain network resource
optimization for a given recovery scheme, for instance (1:1)^n.
However, one has also to consider the resource sharing degree, since
the bandwidth distribution per component Link ID over a given TE
Link is by definition unknown. Therefore, a Maximum Sharing Degree
information can be considered in order to optimize the usage of the
shared resources. In this case and if one defines the shared
recovery bandwidth (in terms bandwidth unit) per TE Link i as r[i],
this implies that the following quantity must be maximized over the
potential candidates: sum {i=1}^N [r{i}/t{i} û b{i}], where N is the
total number of links traversed by a given LSP, t{i} the maximum
reservable bandwidth per TE Link i and b[i] as the sum of the
bandwidth committed for working LSPs and dedicated recovery purposes
per TE Link i. Since b{i} =< t{i}, a fully provisioned TE Link i,
will not be selected during the shared recovery path computation
while a fully reserved TE Link i would result in a ratio of 1.
More generally, one can draw the following mapping between the
available bandwidth at the transport and control plane level:
-------- Max Reservable Bandwidth
-----
-----
-------- Max LSP Bandwidth -------- Max LSP Bandwidth
----- -----
----- <------ b ------> -----
----- -----
----- -----
----- -----
-------- 0 -------- 0
The difference between Max Reservable Bandwidth and the Max LSP
Bandwidth is referred to as the Max Sharable Bandwidth. Within the
quantity, the amount of bandwidth dedicated for shared resource
recovery per TE Link i is defined as r[i] and can be expressed in
terms of component link bandwidth unit.
It has been demonstrated that this Partial Information Routing
approach (also referred to as stochastic approach) can also be
applied to resource shareability given the number of times each SRLG
is protected by a recovery resource, in particular an LSP (see
[BOUILLET]). By flooding this summarized information using a link-
state protocol, recovery path computation and selection for SRLG
diverse recovery paths can be optimized with respect to resource
sharing giving a performance difference of less than 5% compared to
a Full Information Flooding approach (also referred to as
deterministic approach). Note that the stochastic approach can be
D.Papadimitriou et al. - Internet Draft û December 2002 26
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
further extended from the GMPLS signalling applicability viewpoint.
This, by allowing working path-related information (and in
particular, shared recovery bandwidth and SRLG information) to be
exchanged over the recovery LSP in order to enable more efficient
admission control at sharing nodes (as described for instance in
[CCAMP-LI]).
8.5 Summary
One can summarize by the following table the selection of a recovery
scheme/strategy, using the recovery types proposed in [CCAMP-TERM]
and the above discussion.
--------------------------------------------------------------------
| Path Search (computation and selection)
--------------------------------------------------------------------
| Pre-planned | Dynamic
--------------------------------------------------------------------
| | faster recovery | Does not apply
| | less flexible |
| 1 | less robust |
| | most resource consuming |
Path | | |
Setup ---------------------------------------------------------
| | relatively fast recovery | Does not apply
| | relatively flexible |
| 2 | relatively robust |
| | resource consumption |
| | depends on sharing degree |
---------------------------------------------------------
| | relatively fast recovery | less faster (computation)
| | more flexible | most flexible
| 3 | relatively robust | most robust
| | resource consumption | less resource consuming
| | depends on sharing degree |
--------------------------------------------------------------------
1. Path Setup with Resource Reservation (i.e. signalling) and
Selection
2. Path Setup with Resource Reservation (i.e. signalling) w/o
Selection
3. Path Setup w/o Resource Reservation (i.e. signalling) w/o
Selection
As defined in [CCAMP-TERM], the term pre-planned refers to
restoration resource pre-computation, signaling (reservation) and a
priori selection (optional), but not cross-connection.
8.6 Technology Dependence
The above analysis applies in fact to any data oriented circuit
technology with discrete bandwidth increments (like Sonet/SDH, G.709
D.Papadimitriou et al. - Internet Draft û December 2002 27
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
OTN, etc.) being controlled by an IP-centric distributed control
plane.
NOTE: this section is not intended to favor one technology versus
another, it just lists pro and cons for each of them in order to
determine the potential added value of GMPLS-based recovery in their
respective context.
8.6.1 OTN Recovery
OTN Recovery specifics are left for further considerations.
8.6.2 Pre-OTN Recovery
Pre-OTN Recovery specifics (also referred to as ôlambda switchingö)
presents mainly the following advantages:
- benefits from a simpler architecture making it more suitable for
meshed-based recovery schemes (on a per channel basis).
- when providing suppression of intermediate node transponders
implies also that failures (such as LoL) propagates until edge
nodes giving the possibility to initiate upper layer driven
recovery actions.
The main disadvantage comes from the lack of interworking due to the
large amount of failure management (in particular failure
notification protocols) and recovery mechanisms currently available.
8.6.3 Sonet/SDH Recovery
Some of the advantages of the Sonet/SDH and more generically any TDM
layer are:
- Protection schemes are standardized (see [G.841]) and can operate
across protected domains and interwork (see [G.842]).
- Provides failure detection, notification and Automatic Protection
Switching (APS).
- Provides greater control over the granularity of the TDM LPS/Links
that can be recovered with respect to coarser optical channel (or
whole fiber content) recovery switching
Some of the current limitations of the Sonet/SDH layer recovery are:
- Inefficient use of spare capacity: Sonet/SDH protection is largely
applied for ring topologies, where spare capacity often remains
idle, making the efficiency of bandwidth usage an issue.
- Limited topological scope: Use of ring topologies (SNCP or Shared
Protection Rings), reduces the flexibility to deploy somewhat more
complex, but potentially more efficient, mesh-based recovery
D.Papadimitriou et al. - Internet Draft û December 2002 28
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
schemes.
- Lack of traffic priority: as with the optical layer, the SDH/Sonet
layer also cannot distinguish between different priorities of
traffic. For example, it is not possible in SDH or Sonet to switch
EF (Expedited Forwarding) and AF (Assured Forwarding) upper layer
packet flow streams based on priority.
9. Conclusion
TBD.
10. Security Considerations
This document does not introduce or imply any specific security
consideration.
11. References
[BRADNER1] Bradner, S., ôThe Internet Standards Process --
Revision 3ö, BCP 9, RFC 2026, October 1996.
[BRADNER2] Bradner, S., ôKey words for use in RFCs to Indicate
Requirement Levelsö, BCP 14, RFC 2119, March 1997.
[BOUILLET] E.Bouillet et al., ôStochastic Approaches to Compute
Shared Meshed Restored Lightpaths in Optical Network
Architecturesö, INFOCOM 2002, New York City, June 2002.
[CCAMP-LI] G.Li et al. ôRSVP-TE Extensions For Shared-Mesh
Restoration in Transport Networksö, Internet Draft,
Work in progress, draft-li-shared-mesh-restoration-
01.txt, November 2001.
[CCAMP-SRLG] D.Papadimitriou et al., ôShared Risk Link Groups
Encoding and Processing,ö Internet Draft, Work in
progress, draft-papadimitriou-ccamp-srlg-processing-
00.txt, June 2002.
[CCAMP-SRG] S.Dharanikota et al., ôInter domain routing with Shared
Risk Groups,ö Internet Draft, Work in progress,
November 2001.
[CCAMP-TERM] E.Mannie and D.Papadimitriou (Editors), ôRecovery
(Protection and Restoration) Terminology for GMPLS,ö
Internet Draft, Work in progress, draft-mannie-ccamp-
gmpls-recovery-terminology-00.txt, February 2002.
[DEMEESTER] P.Demeester et al., ôResilience in Multilayer
Networksö, IEEE Communications Magazine, Vol. 37, No.
8, August 1998, pp. 70-76.
D.Papadimitriou et al. - Internet Draft û December 2002 29
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
[G.707] ITU-T Recommendation G.707, ôNetwork Node Interface for
the Synchronous Digital Hierarchy (SDH)ö, October 2000.
[G.709] ITU-T Recommendation G.709, ôNetwork Node Interface for
the Optical Transport Network (OTN)ö, February 2001
(and Amendment n1, October 2001).
[G.783] ITU-T Recommendation G.783, ôCharacteristics of
Synchronous Digital Hierarchy (SDH) Equipment
Functional Blocksö
[G.798] ITU-T Recommendation G.798, ôCharacteristics of Optical
Transport Network (OTN) Equipment Functional Blocksö
[G.806] ITU-T Recommendation G.806, ôCharacteristics of
Transport Equipment û Description Methodology and
Generic Functionalityö
[G.826] ITU-T Recommendation G.826, ôPerformance Monitoringö
[G.841] ITU-T Recommendation G.841, ôTypes and Characteristics
of SDH Network Protection Architecturesö
[G.842] ITU-T Recommendation G.842, ôInterworking of SDH
network protection architecturesö
[G.GPS] ITU-T Draft Recommendation G.GPS, Version 2, ôGeneric
Protection Switchingö, Work in progress, May 2002.
[LMP] J.Lang (Editor), ôLink Management Protocol (LMP) v1.0ö
Internet Draft, Work in progress, draft-ietf-ccamp-lmp-
03.txt, February 2002.
[LMP-WDM] A.Fredette and J.Lang (Editors), ôLink Management
Protocol (LMP) for DWDM Optical Line Systems,ö Internet
Draft, Work in progress, draft-ietf-ccamp-lmp-wdm-
00.txt, February 2002.
[MANCHESTER] J.Manchester, P.Bonenfant and C.Newton, ôThe Evolution
of Transport Network Survivability,ö IEEE
Communications Magazine, August 1999.
[MPLS-REC] V.Sharma and F.Hellstrand (Editors) et al., ôA
Framework for MPLS Recoveryö, Internet Draft, Work in
Progress, draft-ietf-mpls-recovery-frmwrk-03.txt, July
2001.
[MPLS-OSU] S.Seetharaman et al, ôIP over Optical Networks: A
Summary of Issuesö, Internet Draft, Work in Progress,
draft-osu-ipo-mpls-issues-02.txt, April 2001.
[TE-NS] K.Owens et al, ôNetwork Survivability Considerations
for Traffic Engineered IP Networksö, Internet Draft,
D.Papadimitriou et al. - Internet Draft û December 2002 30
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
Work in Progress, draft-owens-te-network-survivability-
01.txt, July 2001.
[TE-RH] W.Lai, D.McDysan, J.Boyle, et al, ôNetwork Hierarchy
and Multi-layer Survivabilityö, Internet Draft, Work in
Progress, draft-team-tewg-restore-hierarchy-00.txt,
July 2001.
12. Acknowledgments
The authors would like to thank Fabrice Poppe (Alcatel) and Bart
Rousseau (Alcatel) for their revision effort, Richard Rabbat
(Fujitsu) and David Griffith (NIST) for their useful comments.
13. Author's Addresses
Deborah Brungard (AT&T)
Rm. D1-3C22
200 S. Laurel Ave.
Middletown, NJ 07748, USA
Email: dbrungard@att.com
Sudheer Dharanikota (Nayna)
481 Sycamore Drive
Milpitas, CA 95035, USA
Email: sudheer@nayna.com
Jonathan P. Lang (Calient)
25 Castilian
Goleta, CA 93117, USA
Email: jplang@calient.net
Guangzhi Li (AT&T)
180 Park Avenue,
Florham Park, NJ 07932, USA
Email: gli@research.att.com
Phone: +1 973 360-7376
Eric Mannie (KPNQwest)
Terhulpsesteenweg 6A
1560 Hoeilaart, Belgium
Phone: +32 2 658-5652
Email: eric.mannie@ebone.com
Dimitri Papadimitriou (Alcatel)
Francis Wellesplein, 1
B-2018 Antwerpen, Belgium
Phone: +32 3 240-8491
Email: dimitri.papadimitriou@alcatel.be
Bala Rajagopalan (Tellium)
2 Crescent Place
P.O. Box 901
D.Papadimitriou et al. - Internet Draft û December 2002 31
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
Oceanport, NJ 07757-0901, USA
Phone: +1 732 923-4237
Email: braja@tellium.com
Yakov Rekhter (Juniper)
Email: yakov@juniper.net
D.Papadimitriou et al. - Internet Draft û December 2002 32
draft-papadimitriou-ccamp-gmpls-recovery-analysis-01.txt June 2002
Full Copyright Statement
"Copyright (C) The Internet Society (date). All Rights Reserved.
This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain it
or assist in its implementation may be prepared, copied, published
and distributed, in whole or in part, without restriction of any
kind, provided that the above copyright notice and this paragraph
are included on all such copies and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
followed, or as required to translate it into languages other than
English.
The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on an
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE."
D.Papadimitriou et al. - Internet Draft û December 2002 33