Internet DRAFT - draft-atlas-irs-problem-statement
draft-atlas-irs-problem-statement
Network Working Group A. Atlas, Ed.
Internet-Draft T. Nadeau
Intended status: Informational Juniper Networks
Expires: January 31, 2013 D. Ward
Cisco Systems
July 30, 2012
Interface to the Routing System Problem Statement
draft-atlas-irs-problem-statement-00
Abstract
As modern networks grow in scale and complexity, the need for rapid
and dynamic control increases. With scale, the need to automate even
the simplest operations is important, but even more critical is the
ability to quickly interact with more complex operations such as
policy-based controls.
In order to enable applications to have access to and control over
information in the Internet's routing system, we need a publically
documented interface specification. The interface needs to support
real-time, transaction-based interactions using efficient data models
and encodings. Furthermore, the interface must support a variety of
use cases including those where verified control feed-back loops are
needed.
This document expands upon these statements of requirements to
provide a problem statement for an interface to the Internet routing
system.
Status of this Memo
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 January 31, 2013.
Atlas, et al. Expires January 31, 2013 [Page 1]
Internet-Draft IRS Problem Statement July 2012
Copyright Notice
Copyright (c) 2012 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.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. IRS Model and Problem Area for The IETF . . . . . . . . . . . 3
3. Standard Data-Models of Routing State for Installation . . . . 5
4. Learning Router Information . . . . . . . . . . . . . . . . . 5
5. Desired Aspects of a Protocol for IRS . . . . . . . . . . . . 6
6. Existing Management Interfaces . . . . . . . . . . . . . . . . 7
7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 8
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 8
9. Security Considerations . . . . . . . . . . . . . . . . . . . 8
10. Informative References . . . . . . . . . . . . . . . . . . . . 9
Appendix A. Gaps and Concerns for SNMP . . . . . . . . . . . . . 9
Appendix B. Gaps and Concerns with NetConf . . . . . . . . . . . 10
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 11
Atlas, et al. Expires January 31, 2013 [Page 2]
Internet-Draft IRS Problem Statement July 2012
1. Introduction
As modern networks grow in scale and complexity, the need for rapid
and dynamic control increases. With scale, the need to automate even
the simplest operations is important, but even more critical is the
ability to quickly interact with more complex operations such as
policy-based controls.
With complexity comes the need for more sophisticated automated
applications and orchestration software that can process large
quantities of data, run complex algorithms, and adjust the routing
state as required in order to support the applications, their
calculations and their policies. Changes made to the routing state
of a network by external applications must be verifiable by those
applications to ensure that the correct state has been installed in
the right places.
Mechanisms to support the requirements outlined above have been
developed piecemeal as proprietary solutions to specific situations
and needs. A standard protocol, clear operations that an application
can initiate with that protocol, and data-models to support such
actions would facilitate wide-scale deployment of interoperable
applications and routing systems. That a protocol designed to
facilitate rapid, isolated, secure, and dynamic routing changes is
needed motivates the creation of an Interface to The Routing System
(IRS).
2. IRS Model and Problem Area for The IETF
Managing a network of deployed devices running a variety of routing
protocols involves interactions among multiple different functions
and components that exist within the network. Some of these
components are virtual while some are physical; all should be made
available to be managed and manipulated by applications, given that
appropriate access, authentication, and policy hurdles have been
crossed. The management of only some of these components requires
standardization, as others have already been standardized. The IRS
model is intended to incorporate existing mechanisms where
appropriate, and to build extensions and new protocols where needed.
The IRS model and problem area proposed for IETF work is illustrated
in Figure 1.
Atlas, et al. Expires January 31, 2013 [Page 3]
Internet-Draft IRS Problem Statement July 2012
+***************+ +***************+ +***************+
* Application * * Application * * Application *
+***************+ +***************+ +***************+
^ ^ ^
* * *
* * ****************
* * *
v v v
+---------------+ +---------------+
| IRS Client | | IRS Client |
+---------------+ +---------------+
^ ^
|________________ |
| | <== IRS Protocol
| |
...........................|..|..................................
. v v .
. +*************+ +---------------+ +****************+ .
. * Policy * | | * Routing & * .
. * Database *<***>| IRS Agent |<****>* Signaling * .
. +*************+ | | * Protocols * .
. +---------------+ +****************+ .
. ^ ^ ^ ^ .
. +*************+ * * * * .
. * Topology * * * * * .
. * Database *<*******+ * * v .
. +*************+ * * +****************+ .
. * +********>* RIB Manager * .
. * +****************+ .
. * ^ .
. v * .
. +*******************+ * .
. * Subscription & * * .
. * Configuration * v .
. * Templates for * +****************+ .
. * Measurements, * * FIB Manager * .
. * Events, QoS, etc. * * & Data Plane * .
. +*******************+ +****************+ .
.................................................................
<--> interfaces inside the scope of IRS
+--+ objects inside the scope of IRS
<**> interfaces NOT within the scope of IRS
+**+ objects NOT within the scope of IRS
.... boundary of a router participating in the IRS
Atlas, et al. Expires January 31, 2013 [Page 4]
Internet-Draft IRS Problem Statement July 2012
Figure 1: IRS model and Problem Area
A critical aspect of IRS is defining a suitable protocol or protocols
to carry messages between the IRS Clients and the IRS Agent, and
defining the encapsulation of data within those messages. This
should provide a clear transfer syntax that is straightforward for
applications to use (e.g., a Web Services design paradigm), and
should provide the key features specified in Section 5.
The second critical aspect is semantic-aware data-models for
information in the routing system and in a topology database. The
data-models should be separable across different features of the
managed components, versioned, and combine to provide a network data-
model.
3. Standard Data-Models of Routing State for Installation
There is a need to be able to precisely control routing and signaling
state based upon policy or external measures. This can range from
simple static routes to policy-based routing to static multicast
replication and routing state. This means that the data model
employed needs to handle indirection as well as different types of
tunneling and encapsulation. The relevant MIB modules (for example
[RFC4292]) lack the necessary generality and flexibility. In
addition, by having IRS focus initially on interfaces to the RIB
layer (e.g. RIB, LFIB, multicast RIB, policy-based routing), the
ability to use routing indirection allows flexibility and
functionality that can't be as easily obtained at the forwarding
layer.
Efforts to provide this level of control have focused on
standardizing data models that describe the forwarding plane (e.g.
ForCES [RFC3746]). IRS recognizes that the routing system and a
router's OS provide useful mechanisms that applications could
usefully harness to accomplish application-level goals.
In addition to interfaces to the RIB layer, there is a need to
configure the various routing and signaling protocols with differing
dynamic state based upon application-level policy decisions. The
range desired is not available via MIBs at the present time.
4. Learning Router Information
A router has information that applications may require so that they
can understand the network, verify that programmed state is installed
in the forwarding plane, measure the behavior of various flows, and
Atlas, et al. Expires January 31, 2013 [Page 5]
Internet-Draft IRS Problem Statement July 2012
understand the existing configuration and state of the router. IRS
provides a framework for applications to register for asynchronous
notifications and for them to make specific requests for information.
Although there are efforts to extend the topological information
available, even the best of these (e.g., BGP-LS
[I-D.gredler-idr-ls-distribution]) still provides only the current
active state as seen at the IGP layer and above. Detailed
topological state that provides more information than the current
functional status is needed by applications; only the active paths or
links are known versus those desired or unknown to the routing
topology.
For applications to have a feedback loop that includes awareness of
the relevant traffic, an application must be able to request the
measurement and timely, scalable reporting of data. While a
mechanism such as IPFIX [RFC5470] may be the facilitator for
delivering the data, the need for an application to be able to
dynamically request that measurements be taken and data delivered is
critical.
There are a wide range of events that applications could use for
either verification of router state before other network state is
changed (e.g. that a route has been installed), to act upon changes
to relevant routes by others, or upon router events (e.g. link up/
down). While a few of these (e.g. link up/down) may be available via
MIB Notifications today, the full range is not - nor is there the
ability to set up the router to trigger different actions upon an
event's occurrence.
5. Desired Aspects of a Protocol for IRS
This section describes required aspects of a protocol that could
support IRS. Whether such a protocol is built upon extending
existing mechanisms or requires a new mechanism requires further
investigation.
The key aspects needed in an interface to the routing system are:
Multiple Simultaneous Asynchronous Operations: A single application
should be able to send multiple operations to IRS without needing
to wait for each to complete before sending the next.
Configuration Not Re-Processed: When an IRS operation is processed,
it does not require that any of the configuration be processed.
I.e., the desired behavior is orthogonal to the static
configuration.
Atlas, et al. Expires January 31, 2013 [Page 6]
Internet-Draft IRS Problem Statement July 2012
Duplex: Communications can be established by either the router or
the application. Similarly, events, acknowledgements, failures,
operations, etc. can be sent at any time by both the router and
the application. The IRS is not a pure pull-model where only the
application queries to pull responses.
High-Throughput: At a minimum, the IRS Agent and associated router
should be able to handle hundreds of simple operations per second.
Responsive: It should be possible to complete simple operations
within a sub-second time-scale.
Multi-Channel: It should be possible for information to be
communicated via the interface from different components in the
router without requiring going through a single channel. For
example, for scaling, some exported data or events may be better
sent directly from the forwarding plane, while other interactions
may come from the control-plane. Thus a single TCP session would
not be a good match.
Timing of State Installation and Expiration: The ability to have
state installed with different lifetimes and different start-times
is very valuable. In particular, the ability of an IRS client to
request that a pre-sent operation be started based upon a dynamic
event would provide a powerful functionality.
To extract information in a scalable fashion that is more easily used
by applications, the ability to specify filtering constructs in an
operation requesting data or requesting an asynchronous notification
is very valuable.
6. Existing Management Interfaces
This section discusses the combination of the abstract data models,
their representation in a data language, and the transfer protocol
commonly used with them as a single entity. While other combinations
are possible, the combinations described are those that have
significant deployment.
There are three basic ways that routers are managed. The most
popular is the command line interface (CLI), which allows both
configuration and learning of device state. This is a proprietary
interface resembling a UNIX shell that allows for very customized
control and observation of a device, and, specifically of interest in
this case, its routing system. Some form of this interface exists on
almost every device (virtual or otherwise). Processing of
information returned to the CLI (called "screen scraping") is a
Atlas, et al. Expires January 31, 2013 [Page 7]
Internet-Draft IRS Problem Statement July 2012
burdensome activity because the data is normally formatted for use by
a human operator, and because the layout of the data can vary from
device to device, and between different software versions. Despite
its ubiquity, this interface has never been standardized and is
unlikely to ever be standardized. IRS does not involve CLI
standardization.
The second most popular interface for interrogation of a device's
state, statistics, and configuration is The Simple Network Management
Protocol (SNMP) and a set of relevant standards-based and proprietary
Management Information Base (MIB) modules. SNMP has a strong history
of being used by network managers to gather statistical and state
information about devices, including their routing systems. However,
SNMP is very rarely used to configure a device or any of its systems
for reasons that vary depending upon the network operator. Some
example reasons include complexity, the lack of desired configuration
semantics (e.g., configuration "roll-back", "sandboxing" or
configuration versioning), and the difficulty of using the semantics
(or lack thereof) as defined in the MIB modules to configure device
features. Therefore, SNMP is not considered as a candidate solution
for the problems motivating IRS.
Finally, the IETF's Network Configuration (or NetConf) protocol has
made many strides at overcoming most of the limitations around
configuration that were just described. However, the lack of
standard data models have hampered the adoption of NetConf.
7. Acknowledgements
The authors would like to thank Ken Gray for their suggestions and
review.
8. IANA Considerations
This document includes no request to IANA.
9. Security Considerations
Security is a key aspect of any protocol that allows state
installation and extracting of detailed router state. More
investigation remains to fully define the security requirements, such
as authorization and authentication levels.
Atlas, et al. Expires January 31, 2013 [Page 8]
Internet-Draft IRS Problem Statement July 2012
10. Informative References
[I-D.gredler-idr-ls-distribution]
Gredler, H., Medved, J., Previdi, S., and A. Farrel,
"North-Bound Distribution of Link-State and TE Information
using BGP", draft-gredler-idr-ls-distribution-02 (work in
progress), July 2012.
[RFC3746] Yang, L., Dantu, R., Anderson, T., and R. Gopal,
"Forwarding and Control Element Separation (ForCES)
Framework", RFC 3746, April 2004.
[RFC4292] Haberman, B., "IP Forwarding Table MIB", RFC 4292,
April 2006.
[RFC5470] Sadasivan, G., Brownlee, N., Claise, B., and J. Quittek,
"Architecture for IP Flow Information Export", RFC 5470,
March 2009.
Appendix A. Gaps and Concerns for SNMP
Though SNMP can allow state to be written, the overhead of the
required infrastructure is quite high. Clients and servers that wish
to use SNMP must build in and understand a large number of MIB
modules, including many proprietary modules. Even when ignoring the
overhead in building the SNMP processing and handling functions into
an application, these properties lend themselves well to read-only
operations. A critical lack in MIB modules for read-write (i.e., for
configuration) operations is that there is no semantic understanding
of the objects defined in the modules encoded in those modules. Any
required semantic knowledge must be specifically hand-coded into
applications or ignored. Further, many MIB modules do not allow the
writing of state, and this limits coverage; owing to the cumbersome
nature, there has not been interest in increasing coverage.
An additional deficiency in using SNMP MIB modules either for reading
or writing comes in the inherent co-mingling of semantics and syntax
in the form of indexing requirements. SNMP MIB modules contain
tables that also define an index format. This format is then
translated - often mapped onto - a device's actual implementation.
Such a mapping means an implementation either matches the module's
indexing during searches or not; if not, then an implementation is
slowed down when it searches for objects. Even in not-so-extreme
cases, such slow performance can result in the SNMP manager's request
timing-out owing to the delay of the SNMP agent's response.
For example, if a search of N*M objects is stipulated as (N, M) in
Atlas, et al. Expires January 31, 2013 [Page 9]
Internet-Draft IRS Problem Statement July 2012
the standard MIB module, but the implementation happens to choose to
index its tables internally as (M, N), this will result in search
times of O(N^2). When N or M become large, as they do in routing
tables, this results in wasted processing cycles for the device, and
either extremely long wait times for queries, or simply a lack of
answers to queries. It is a clear requirement for the IRS to not
suffer from this issue.
In addition to these difficulties, SNMP matches up to the key needed
aspects as follows:
Multiple Simultaneous Asynchronous Operations: Supported, but
difficult for configuration.
Configuration Not Re-Processed: Supported
Duplex: The manager must initiate communications with the SNMP
agent on the router. With the limited exception of Notifications
and InformRequests defined in a MIB module, this is a pull model.
High-Throughput: Possible
Responsive: Possible
Multi-Channel: Possible
The key gaps identified for SNMP are:
a. Infrastructure Overhead
b. Lack of Semantic Information in the Data-Model
c. Required Indexing, from mingling of semantics and syntax, badly
impacting performance or driving implementation decisions.
d. Limited interest and use for configuration
e. Communication model isn't naturally duplex.
Appendix B. Gaps and Concerns with NetConf
While NetConf has solved many of the deficiencies present in SNMP in
terms of configuration, it still does not satisfy a number of
requirements needed to manage today's routing information. First,
the lack of standard data models have hampered the adoption of
NetConf; a significant amount of per-vendor customization is still
needed. The transport mechanisms that are currently defined (e.g.,
Atlas, et al. Expires January 31, 2013 [Page 10]
Internet-Draft IRS Problem Statement July 2012
SOAP/BEEP) for NetConf are not those commonly used by modern
applications (e.g., ReST or JSON).
NetConf primarily facilitates configuration rather than reading of
state or handling asynchronous events.
NetConf matches up to the key needed aspects as follows:
Multiple Simultaneous Asynchronous Operations: Not Possible
Configuration Not Re-Processed: Not Possible
Duplex: Not Possible - strict pull model.
High-Throughput: Unlikely - Can depend on configuration size
Responsive: Unlikely - Can depend on configuration size
Multi-Channel: Not Possible
Authors' Addresses
Alia Atlas (editor)
Juniper Networks
10 Technology Park Drive
Westford, MA 01886
USA
Email: akatlas@juniper.net
Thomas Nadeau
Juniper Networks
1194 N. Mathilda Ave.
Sunnyvale, CA 94089
USA
Email: tnadeau@juniper.net
Atlas, et al. Expires January 31, 2013 [Page 11]
Internet-Draft IRS Problem Statement July 2012
Dave Ward
Cisco Systems
Tasman Drive
San Jose, CA 95134
USA
Email: wardd@cisco.com
Atlas, et al. Expires January 31, 2013 [Page 12]
