| Network Working Group | Y. Nishida |
| Internet-Draft | GE Global Research |
| Intended status: Standards Track | P. Natarajan |
| Expires: August 20, 2016 | Cisco Systems |
| A. Caro | |
| BBN Technologies | |
| P. Amer | |
| University of Delaware | |
| K. Nielsen | |
| Ericsson | |
| February 17, 2016 |
SCTP-PF: Quick Failover Algorithm in SCTP
draft-ietf-tsvwg-sctp-failover-16.txt
SCTP supports multi-homing. However, when the failover operation specified in RFC4960 is followed, there can be significant delay and performance degradation in the data transfer path failover. To overcome this problem this document specifies a quick failover algorithm (SCTP-PF) based on the introduction of a Potentially Failed (PF) state in SCTP Path Management.
The document also specifies a dormant state operation of SCTP. This dormant state operation is required to be followed by an SCTP-PF implementation, but it may equally well be applied by a standard RFC4960 SCTP implementation.
Additionally, the document introduces an alternative switchback operation mode called Primary Path Switchover that will be beneficial in certain situations. This mode of operation applies to both a standard RFC4960 SCTP implementation as well as to a SCTP-PF implementation.
The procedures defined in the document require only minimal modifications to the RFC4960 specification. The procedures are sender-side only and do not impact the SCTP receiver.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
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."
This Internet-Draft will expire on August 20, 2016.
Copyright (c) 2016 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
The Stream Control Transmission Protocol (SCTP) specified in [RFC4960] supports multi-homing at the transport layer. SCTP's multi-homing features include failure detection and failover procedures to provide network interface redundancy and improved end-to-end fault tolerance. In SCTP's current failure detection procedure, the sender must experience Path.Max.Retrans (PMR) number of consecutive failed timer-based retransmissions on a destination address before detecting a path failure. Until detecting the path failure, the sender continues to transmit data on the failed path. The prolonged time in which [RFC4960] SCTP continues to use a failed path severely degrades the performance of the protocol. To address this problem, this document specifies a quick failover algorithm (SCTP-PF) based on the introduction of a new Potentially Failed (PF) path state in SCTP path management. The performance deficiencies of the [RFC4960] failover operation, and the improvements obtainable from the introduction of a Potentially Failed state in SCTP, were proposed and documented in [NATARAJAN09] for Concurrent Multipath Transfer SCTP [IYENGAR06].
While SCTP-PF can accelerate failover process and improve performance, the risks that an SCTP endpoint enters the dormant state where all destination addresses are inactive can be increased. [RFC4960] leaves the protocol operation during dormant state to implementations and encourages to avoid entering the state as much as possible by careful tuning of the Path.Max.Retrans (PMR) and Association.Max.Retrans (AMR) parameters. We specify a dormant state operation for SCTP-PF which makes SCTP-PF provide the same disruption tolerance as [RFC4960] despite that the dormant state may be entered more quickly. The dormant state operation may equally well be applied by an [RFC4960] implementation and will here serve to provide added fault tolerance for situations where the tuning of the Path.Max.Retrans (PMR) and Association.Max.Retrans (AMR) parameters fail to provide adequate prevention of the entering of the dormant state.
The operation after the recovery of a failed path also impacts the performance of the protocol. With the procedures specified in [RFC4960] SCTP will, after a failover from the primary path, switch back to use the primary path for data transfer as soon as this path becomes available again. From a performance perspective such a forced switchback of the data transmission path can be suboptimal as the CWND towards the original primary destination address has to be rebuilt once data transfer resumes, [CARO02]. As an optional alternative to the switchback operation of [RFC4960], this document specifies an alternative Primary Path Switchover procedure which avoid such forced switchbacks of the data transfer path. The Primary Path Switchover operation was originally proposed in [CARO02].
While SCTP-PF primarily is motivated by a desire to improve the multi-homed operation, the feature applies also to SCTP single-homed operation. Here the algorithm serves to provide increased failure detection on idle associations, whereas the failover or switchback aspects of the algorithm will not be activated. This is discussed in more detail in Appendix C.
A brief description of the motivation for the introduction of the Potentially Failed state including a discussion of alternative approaches to mitigate the deficiencies of the [RFC4960] failover operation are given in the Appendices. Discussion of path bouncing effects that might be caused by frequent switchovers, are also provided there.
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 [RFC2119].
To minimize the performance impact during failover, the sender should avoid transmitting data to a failed destination address as early as possible. In the [RFC4960] SCTP path management scheme, the sender stops transmitting data to a destination address only after the destination address is marked inactive. This process takes a significant amount of time as it requires the error counter of the destination address to exceed the Path.Max.Retrans (PMR) threshold. The issue cannot simply be mitigated by lowering of the PMR threshold because this may result in spurious failure detection and unnecessary prevention of the usage of a preferred primary path. Also due to the coupled tuning of the Path.Max.Retrans (PMR) and the Association.Max.Retrans (AMR) parameter values in [RFC4960], lowering of the PMR threshold may result in lowering of the AMR threshold, which would result in decrease of the fault tolerance of SCTP.
The solution provided in this document is to extend the SCTP path management scheme of [RFC4960] by the addition of the Potentially Failed (PF) state as an intermediate state in between the active and inactive state of a destination address in the [RFC4960] path management scheme, and let the failover of data transfer away from a destination address be driven by the entering of the PF state instead of by the entering of the inactive state. Thereby SCTP may perform quick failover without negatively impacting the overall fault tolerance of [RFC4960] SCTP. At the same time, RTO-based HEARTBEAT probing is initiated towards a destination address once it enters PF state. Thereby SCTP may quickly ascertain whether network connectivity towards the destination address is broken or whether the failover was spurious. In the case where the failover was spurious data transfer may quickly resume towards the original destination address.
The new failure detection algorithm assumes that loss detected by a timeout implies either severe congestion or network connectivity failure. It recommends that by default a destination address is classified as PF at the occurrence of the first timeout.
The SCTP-PF operation is specified as follows:
When choosing among multiple destination addresses in active state an SCTP sender will follow the guiding principles of section 6.4.1 of
[RFC4960] of choosing most divergent source-destination pairs compared with, for i.: the destination address in PF state that it performs a failover from, and for ii.: the destination address towards which the data timed out. Rules for picking the most divergent source-destination pair are an implementation decision and are not specified within this document.The sender MUST NOT change the state and the error counter of any destination addresses as the result of the selection.
In a situation with complete disruption of the communication in between the SCTP Endpoints, the aggressive HEARTBEAT transmissions of SCTP-PF on destination addresses in PF state may make the association enter dormant state faster than a standard [RFC4960] SCTP implementation given the same setting of Path.Max.Retrans (PMR) and Association.Max.Retrans (AMR). For example, an SCTP association with two destination addresses typically would reach dormant state in half the time of an [RFC4960] SCTP implementation in such situations. This is because a SCTP PF sender will send HEARTBEATS and data retransmissions in parallel with RTO intervals when there are multiple destinations addresses in PF state. This argument presumes that RTO << HB.interval of [RFC4960]. With the design goal that SCTP-PF shall provide the same level of disruption tolerance as an [RFC4960] SCTP implementation with the same Path.Max.Retrans (PMR) and Association.Max.Retrans (AMR) setting, we prescribe for that an SCTP-PF implementation SHOULD operate as described below in Section 4.1 during dormant state.
An SCTP-PF implementation MAY choose a different dormant state operation than the one described below in Section 4.1 provided that the solution chosen does not decrease the fault tolerance of the SCTP-PF operation.
The below prescription for SCTP-PF dormant state handling MUST NOT be coupled to the value of the PFMR, but solely to the activation of SCTP-PF logic in an SCTP implementation.
It is noted that the below dormant state operation is considered to provide added disruption tolerance also for an [RFC4960] SCTP implementation, and that it can be sensible for an [RFC4960] SCTP implementation to follow this mode of operation. For an [RFC4960] SCTP implementation the continuation of data transmission during dormant state makes the fault tolerance of SCTP be more robust towards situations where some, or all, alternative paths of an SCTP association approach, or reach, inactive state before the primary path used for data transmission observes trouble.
The objective of the Primary Path Switchover operation is to allow the SCTP sender to continue data transmission on a new working path even when the old primary destination address becomes active again. This is achieved by having SCTP perform a switchover of the primary path to the new working path if the error counter of the primary path exceeds a certain threshold. This mode of operation can be applied not only to SCTP-PF implementations, but also to [RFC4960] implementations.
The Primary Path Switchover operation requires only sender side changes. The details are:
The manner of switchover operation that is most optimal in a given scenario depends on the relative quality of a set primary path versus the quality of alternative paths available as well as on the extent to which it is desired for the mode of operation to enforce traffic distribution over a number of network paths. I.e., load distribution of traffic from multiple SCTP associations may be sought to be enforced by distribution of the set primary paths with [RFC4960] switchback operation. However as [RFC4960] switchback behavior is suboptimal in certain situations, especially in scenarios where a number of equally good paths are available, an SCTP implementation MAY support also, as alternative behavior, the Primary Path Switchover mode of operation and MAY enable it based on applications' requests.
For an SCTP implementation that implements the Primary Path Switchover operation, this specification RECOMMENDS that the standard RFC4960 switchback operation is retained as the default operation.
This document does not alter the [RFC4960] value recommendation for the SCTP Protocol Parameters defined in [RFC4960].
The following protocol parameter is RECOMMENDED:
This section describes how the socket API defined in [RFC6458] is extended to provide a way for the application to control and observe the SCTP-PF behavior as well as the Primary Path Switchover function.
Please note that this section is informational only.
A socket API implementation based on [RFC6458] is, by means of the existing SCTP_PEER_ADDR_CHANGE event, extended to provide the event notification when a peer address enters or leaves the potentially failed state as well as the socket API implementation is extended to expose the potentially failed state of a peer address in the existing SCTP_GET_PEER_ADDR_INFO structure.
Furthermore, two new read/write socket options for the level IPPROTO_SCTP and the name SCTP_PEER_ADDR_THLDS and SCTP_EXPOSE_POTENTIALLY_FAILED_STATE are defined as described below. The first socket option is used to control the values of the PFMR and PSMR parameters described in Section 3 and in Section 5. The second one controls the exposition of the potentially failed path state.
Support for the SCTP_PEER_ADDR_THLDS and SCTP_EXPOSE_POTENTIALLY_FAILED_STATE socket options need also to be added to the function sctp_opt_info().
As defined in [RFC6458], the SCTP_PEER_ADDR_CHANGE event is provided if the status of a peer address changes. In addition to the state changes described in [RFC6458], this event is also provided, if a peer address enters or leaves the potentially failed state. The notification as defined in [RFC6458] uses the following structure:
struct sctp_paddr_change {
uint16_t spc_type;
uint16_t spc_flags;
uint32_t spc_length;
struct sockaddr_storage spc_aaddr;
uint32_t spc_state;
uint32_t spc_error;
sctp_assoc_t spc_assoc_id;
}
[RFC6458] defines the constants SCTP_ADDR_AVAILABLE, SCTP_ADDR_UNREACHABLE, SCTP_ADDR_REMOVED, SCTP_ADDR_ADDED, and SCTP_ADDR_MADE_PRIM to be provided in the spc_state field. This document defines in addition to that the new constant SCTP_ADDR_POTENTIALLY_FAILED, which is reported if the affected address becomes potentially failed.
The SCTP_GET_PEER_ADDR_INFO socket option defined in [RFC6458] can be used to query the state of a peer address. It uses the following structure:
struct sctp_paddrinfo {
sctp_assoc_t spinfo_assoc_id;
struct sockaddr_storage spinfo_address;
int32_t spinfo_state;
uint32_t spinfo_cwnd;
uint32_t spinfo_srtt;
uint32_t spinfo_rto;
uint32_t spinfo_mtu;
};
[RFC6458] defines the constants SCTP_UNCONFIRMED, SCTP_ACTIVE, and SCTP_INACTIVE to be provided in the spinfo_state field. This document defines in addition to that the new constant SCTP_POTENTIALLY_FAILED, which is reported if the peer address is potentially failed.
Applications can control the SCTP-PF behavior by getting or setting the number of consecutive timeouts before a peer address is considered potentially failed or unreachable. The same socket option is used by applications to set and get the number of timeouts before the primary path is changed automatically by the Primary Path Switchover function. This socket option uses the level IPPROTO_SCTP and the name SCTP_PEER_ADDR_THLDS.
The following structure is used to access and modify the thresholds:
struct sctp_paddrthlds {
sctp_assoc_t spt_assoc_id;
struct sockaddr_storage spt_address;
uint16_t spt_pathmaxrxt;
uint16_t spt_pathpfthld;
uint16_t spt_pathcpthld;
};
Applications can control the exposure of the potentially failed path state in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO as described in Section 7.1. The default value is implementation specific.
This socket option uses the level IPPROTO_SCTP and the name SCTP_EXPOSE_POTENTIALLY_FAILED_STATE.
The following structure is used to control the exposition of the potentially failed path state:
struct sctp_assoc_value {
sctp_assoc_t assoc_id;
uint32_t assoc_value;
};
Security considerations for the use of SCTP and its APIs are discussed in [RFC4960] and [RFC6458].
The logic introduced by this document does not impact existing SCTP messages on the wire. Also, this document does not introduce any new SCTP messages on the wire that require new security considerations.
SCTP-PF makes SCTP not only more robust during primary path failure/congestion but also more vulnerable to network connectivity/congestion attacks on the primary path. SCTP-PF makes it easier for an attacker to trick SCTP to change data transfer path, since the duration of time that an attacker needs to negatively influence the network connectivity is much shorter than [RFC4960]. However, SCTP-PF does not constitute a significant change in the duration of time and effort an attacker needs to keep SCTP away from the primary path. With the standard switchback operation [RFC4960] SCTP resumes data transfer on its primary path as soon as the next HEARTBEAT succeeds.
On the other hand, usage of the Primary Path Switchover mechanism, does change the threat analysis. This is because on-path attackers can force a permanent change of the data transfer path by blocking the primary path until the switchover of the primary path is triggered by the Primary Path Switchover algorithm. This especially will be the case when the Primary Path Switchover is used together with SCTP-PF with the particular setting of PSMR = PFMR = 0, as Primary Path Switchover here happens already at the first RTO timeout experienced. Users of the Primary Path Switchover mechanism should be aware of this fact.
The event notification of path state transfer from active to potentially failed state and vice versa gives attackers an increased possibility to generate more local events. However, it is assumed that event notifications are rate-limited in the implementation to address this threat.
SCTP-PF introduces new SCTP algorithms for failover and switchback with associated new state parameters. It is recommended that the SCTP-MIB defined in [RFC3873] is updated to support the management of the SCTP-PF implementation. This can be done by extending the sctpAssocRemAddrActive field of the SCTPAssocRemAddrTable to include information of the PF state of the destination address and by adding new fields to the SCTPAssocRemAddrTable supporting PotentiallyFailed.Max.Retrans (PFMR) and Primary.Switchover.Max.Retrans (PSMR) parameters.
This document does not create any new registries or modify the rules for any existing registries managed by IANA.
The authors wish to thank Michael Tuexen for his many invaluable comments and for his very substantial support with the making of this document.
Initially this work looked to entail some changes of the Congestion Control (CC) operation of SCTP and for this reason the work was proposed as Experimental. These intended changes of the CC operation have since been judged to be irrelevant and are no longer part of the specification. As the specification entails no other potential harmful features, consensus exists in the WG to bring the work forward as PS.
Initially concerns have been expressed about the possibility for the mechanism to introduce path bouncing with potential harmful network impacts. These concerns are believed to be unfounded. This issue is addressed in Appendix B.
It is noted that the feature specified by this document is implemented by multiple SCTP SW implementations and furthermore that various variants of the solution have been deployed in telephony signaling environments for several years with good results.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC4960] | Stewart, R., "Stream Control Transmission Protocol", RFC 4960, DOI 10.17487/RFC4960, September 2007. |
This section lists alternative approaches for the issues described in this document. Although these approaches do not require to update RFC4960, we do not recommend them from the reasons described below.
Smaller values for Path.Max.Retrans shorten the failover duration and in fact this is recommended in some research results [JUNGMAIER02] [GRINNEMO04] [FALLON08]. However to significantly reduce the failover time it is required to go down (as with PFMR) to Path.Max.Retrans=0 and with this setting SCTP switches to another destination address already on a single timeout which may result in spurious failover. Spurious failover is a problem in [RFC4960] SCTP as the transmission of HEARTBEATS on the left primary path, unlike in SCTP-PF, is governed by 'HB.interval' also during the failover process. 'HB.interval' is usually set in the order of seconds (recommended value is 30 seconds) and when the primary path becomes inactive, the next HEARTBEAT may be transmitted only many seconds later. Indeed as recommended, only 30 secs later. Meanwhile, the primary path may since long have recovered, if it needed recovery at all (indeed the failover could be truly spurious). In such situations, post failover, an endpoint is forced to wait in the order of many seconds before the endpoint can resume transmission on the primary path and furthermore once it returns on the primary path the CWND needs to be rebuild anew - a process which the throughput already have had to suffer from on the alternate path. Using a smaller value for 'HB.interval' might help this situation, but it would result in a general waste of bandwidth as such more frequent HEARTBEATING would take place also when there are no observed troubles. The bandwidth overhead may be diminished by having the ULP use a smaller 'HB.interval' only on the path which at any given time is set to be the primary path, but this adds complication in the ULP.
In addition, smaller Path.Max.Retrans values also affect the 'Association.Max.Retrans' value. When the SCTP association's error count exceeds Association.Max.Retrans threshold, the SCTP sender considers the peer endpoint unreachable and terminates the association. Section 8.2 in [RFC4960] recommends that Association.Max.Retrans value should not be larger than the summation of the Path.Max.Retrans of each of the destination addresses. Else the SCTP sender considers its peer reachable even when all destinations are INACTIVE and to avoid this dormant state operation, [RFC4960] SCTP implementation SHOULD reduce Association.Max.Retrans accordingly whenever it reduces Path.Max.Retrans. However, smaller Association.Max.Retrans value decreases the fault tolerance of SCTP as it increases the chances of association termination during minor congestion events.
As several research results indicate, we can also shorten the duration of failover process by adjusting RTO related parameters [JUNGMAIER02] [FALLON08]. During failover process, RTO keeps being doubled. However, if we can choose smaller value for RTO.max, we can stop the exponential growth of RTO at some point. Also, choosing smaller values for RTO.initial or RTO.min can contribute to keep the RTO value small.
Similar to reducing Path.Max.Retrans, the advantage of this approach is that it requires no modification to the current specification, although it needs to ignore several recommendations described in the Section 15 of [RFC4960]. However, this approach requires to have enough knowledge about the network characteristics between end points. Otherwise, it can introduce adverse side-effects such as spurious timeouts.
The significant issue with this approach, however, is that even if the RTO.max is lowered to an optimal low value, then as long as the Path.Max.Retrans is kept at the [RFC4960] recommended value, the reduction of the RTO.max doesn't reduce the failover time sufficiently enough to prevent severe performance degradation during failover.
The methods described in the document can accelerate the failover process. Hence, they might introduce the path bouncing effect where the sender keeps changing the data transmission path frequently. This sounds harmful to the data transfer, however several research results indicate that there is no serious problem with SCTP in terms of path bouncing effect [CARO04] [CARO05].
There are two main reasons for this. First, SCTP is basically designed for multipath communication, which means SCTP maintains all path related parameters (CWND, ssthresh, RTT, error count, etc) per each destination address. These parameters cannot be affected by path bouncing. In addition, when SCTP migrates the data transfer to another path, it starts with the minimal or the initial CWND. Hence, there is little chance for packet reordering or duplicating.
Second, even if all communication paths between the end-nodes share the same bottleneck, the SCTP-PF results in a behavior already allowed by [RFC4960].
For a single-homed SCTP association the only tangible effect of the activation of SCTP-PF operation is enhanced failure detection in terms of potential notification of the PF state of the sole destination address as well as, for idle associations, more rapid entering, and notification, of inactive state of the destination address and more rapid end-point failure detection. It is believed that neither of these effects are harmful, provided adequate dormant state operation is implemented, and furthermore that they may be particularly useful for applications that deploys multiple SCTP associations for load balancing purposes. The early notification of the PF state may be used for preventive measures as the entering of the PF state can be used as a warning of potential congestion. Depending on the PMR value, the aggressive HEARTBEAT transmission in PF state may speed up the end-point failure detection (exceed of AMR threshold on the sole path error counter) on idle associations in case where relatively large HB.interval value compared to RTO (e.g. 30secs) is used.