| ICNRG | A. Rahman |
| Internet-Draft | D. Trossen |
| Intended status: Informational | InterDigital Inc. |
| Expires: December 31, 2017 | D. Kutscher |
| R. Ravindran | |
| Huawei | |
| June 29, 2017 |
Deployment Considerations for Information-Centric Networking (ICN)
draft-rahman-icnrg-deployment-guidelines-02
Information-Centric Networking (ICN) is now reaching technological maturity after many years of fundamental research and experimentation. This document provides a number of deployment considerations in the interest of helping the ICN community move forward to the next step of live deployments. The major deployment configurations for ICN are first described including the main overlay and underlay approaches. Then proposed deployment migration paths are outlined to address major practical issues such as network and application migration. Next, selected ICN trial experiences are summarized. Finally, protocol areas that require further standardization are identified to facilitate future interoperable ICN deployments.
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 December 31, 2017.
Copyright (c) 2017 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 ICNRG charter identifies deployment guidelines as an important topic area for the ICN community. Specifically, the charter states that defining concrete migration paths for ICN deployments which avoid forklift upgrades, and defining practical ICN interworking configurations with the existing Internet paradigm, are key topic areas that require further investigation. Also, it is well understood that results and conclusions from any mid to large-scale ICN experiments in the live Internet will also provide useful guidance for deployments.
However, so far outside of some preliminary investigations such as [I-D.paik-icn-deployment-considerations], there has not been much progress on this topic. This draft attempts to fill some of these gaps by defining clear deployment configurations for ICN, and associated migration pathways for these configurations. Also, selected deployment trial experiences of ICN technology are summarized. Finally, recommendations are made for potential future IETF standardization of key protocol functionality that will facilitate interoperable ICN deployments going forward.
This document assumes readers are, in general, familiar with the terms and concepts that are defined in [RFC7927]. In addition, this document defines the following terminology:
In this section, we present various deployment options for ICN. These are presented as "configurations" that allow for studying these options further. While this document will outline experiences with various of these configurations (in Section 5), we will not provide an in-depth technical or commercial evaluation for any of them - for this we refer to existing literature in this space such as [Tateson].
ICN has often been described as a "clean-slate" approach with the goal to renew or replace the current IP routing fabric of the Internet. As such, existing routing hardware as well as ancillary services are not taken for granted. This clean-slate view is reflected as deployment configurations we label as "wholesale replacement" of large part of the existing Internet infrastructure. For instance, such configuration would see existing IP routers being replaced by ICN-specific forwarding and routing elements, such as NFD (Named Data Networking Forwarding Daemon) [NFD], CCN routers [Jacobson] or PURSUIT forwarding nodes [IEEE_Communications]. All major ICN approaches have explored this option as one of their paths to deployment.
While such replacement could be seen as exclusive for ICN deployments, some ICN approaches [POINT] rely on the deployment of infrastructure upgrades, here SDN switches. Such upgrades, while being possibly utilized for a "clean slate" ICN deployment would not necessary be used exclusively for an ICN deployment. Different proposals have been made for various ICN approaches to enable the operation over an SDN transport [Reed][CONET][C_FLOW].
Similar to other significant changes to the Internet routing fabric, particularly the transition from IPv4 to IPv6 or the introduction of IP multicast, this deployment configuration foresees the creation of an ICN overlay. Note that this overlay approach is sometimes, informally, also referred to as a tunneling approach. The overlay approach could be done directly such as ICN-over-UDP as described in [CCNx_UDP]. Alternatively, the overlay could be done via ICN-in-L2-in-IP as in [IEEE_Communications] which describes a recursive layering process.
Another flavor of overlay would be embedding ICN semantics into existing protocols. A recently announced approach is [Hybrid_ICN] where the ICN names are mapped to IPv6 addresses. Another approach used in the Network of Information (NetInf) is to define a convergence layer to map NetInf semantics to HTTP [I-D.kutscher-icnrg-netinf-proto]. Regardless of the flavor, however, the overlay approach results in islands of ICN deployments over existing IP-based infrastructure.
Proposals such as [POINT] and [White] outline the deployment option of using an ICN underlay that would integrate with existing IP-based services by deploying application layer gateways at appropriate locations, while possibly still allowing for native ICN applications to emerge. The main reasons for such configuration option is the backward-compatible introduction of ICN technology, while reaping the benefits of ICN in terms of multicast delivery, mobility support, fast indirection due to location independence, in-network computing and possibly more.
In this sub-option, a core network would utilize edge-based protocol mapping onto the native ICN underlay. For instance, [POINT] proposes to map HTTP transactions, or some other IP based transactions such as CoAP, directly onto an ICN-based message exchange. This mapping is realized at the network attachment point, such as realized in access points or customer premise equipment, which in turn provides a standard IP interface to existing user devices. Towards peering networks, such network attachment point turns into a modified border gateway/proxy, preserving the perception of an IP-based core network towards any peering network.
The work in [White] proposes a similar deployment configuration. Here, the target is the use of ICN for content distribution within CDN server farms, i.e., the protocol mapping is realized at the ingress of the server farm where the HTTP-based retrieval request is served, while the response is delivered through a suitable egress node translation.
Native ICN networks may be found at the edge of the network, mostly proposed for Internet of Things (IoT) deployments, which allows the possibility of introducing new network architectures and protocols, and in this context ICN can be a possible candidate considering its suitability for IoT and fixed network scenarios [I-D.zhang-icnrg-icniot]. The integration with the current IP protocol suite takes place at an application gateway/proxy at the edge network boundary, e.g., translating incoming CoAP request/response transactions [RFC7252] into ICN message exchanges or vice versa. Furthermore, ICN will allow us to evolve the role of gateways/proxies as ICN message security should be preserved through the protocol translation function of a gateway/proxy and thus offer a substantial gain.
The work in [VSER] positions ICN as an edge service gateway driven by an generalized ICN based service orchestration system with its own compute and network virtualization controllers to manage an ICN infrastructure. The platform also offers service discovery capabilities to enable user applications to discover appropriate ICN service gateways. To exemplify a use case scenario, the platform shows the realization of a multi-party audio/video conferencing service over such a edge cloud deployment of ICN routers realized over commodity hardware platforms. This platform has also been extended to offer seamless mobility and mobility as a service [VSER-Mob] features.
The objective of Network slicing [NGMN]is to multiplex a general pool of compute, storage and bandwidth resources among multiple services with exclusive SLA requirements on transport level QoS and security. These services could include both connectivity services like LTE-as-a-service or OTT services like VoD or other IoT services. Such a framework can also be used to realize ICN slices with its own control and forwarding plane over which one or more end-user services can be delivered.
An ICN slice can itself be overlaid over IP or can be underlaid using generalized programmable data planes like P4/POF. Such a generalized network slicing framework should be able to offer service slices to be realized over both IP and ICN. Network slicing will rely heavily on network softwarization and programmability using SDN/NFV technologies for efficient utilization of available resources without compromising on the slice requirements. Coupled with the view of ICN functions as being "chained service functions" [RFC7665], an ICN deployment within such a slice could be realized within the emerging control plane that is targeted for adoption in future (e.g., 5th generation mobile) network deployments. Finally, it should be noted that ICN is not creating the network slice, but instead that the slice is created to run a 5G-ICN instance [Ravindran].
At the level of the specific technologies involved, the 5G-ICN slice requires compatibility for instance at the level of the forwarding/data plane depending on if it is realized as an overlay or using programmable data planes. With SDN emerging for new network deployments, some ICN approaches will need to integrate with SDN as a data plane forwarding function, as briefly discussed in Section 3.1.
After outlining the various ICN deployment configurations in Section 3, we now focus on the various migration paths that will have importance to the various stakeholders that are usually involved in the deployment of a technology at (ultimately) large scale. We can identify these stakeholders as: [Tateson], [Techno_Economic] and [Internet_Pricing] are left out of our discussion.
Note that our presentation purely focuses on technological aspects of such migration. Economic or regulatory aspects, such as studied in
The internet is full of applications and services, utilizing the innovation capabilities of the many protocols defined over the packet level IP service. HTTP provides one convergence point for these services with many web development frameworks based on the semantics provided by the hypertext transfer protocol. In recent years, even services such as video delivery have been migrating from the traditional RTP-over-UDP delivery to the various HTTP-level streaming solutions, such as DASH [DASH] and others. Nonetheless, many non-HTTP services exist, all of which need consideration when migrating from the IP-based internet to an ICN-based one.
The underlay deployment configuration options presented in Section 3.3.1 and Section 3.3.2 aim at providing some level of backward compatibility to this existing ecosystem through a proxy based message flow mapping mechanism (e.g., mapping of existing HTTP/TCP/IP message flows to HTTP/TCP/IP/ICN message flows). A related approach of mapping TCP/IP to TCP/ICN message flows is described in [Moiseenko]
Alternatively, ICN as an overlay (Section 3.2), as well as ICN-as-a-Slice (Section 3.4), allow for the introduction of the full capabilities of ICN through new service interfaces as well as operations in the network. With that, these approaches of deployment are likely to aim at introducing new services capitalizing on those ICN capabilities.
A significant number of services and applications are devoted to content delivery in some form, either as video delivery services, social media platforms, and many others. Content delivery networks (CDNs) are deployed to assist these services through localizing the content requests and therefore reducing latency and possibly increase utilization of available bandwidth as well as reducing the load on origin servers. Similar to the previous sub-section, the underlay deployment configurations presented in Section 3.3.1 and Section 3.3.2 aim at providing a migration path for existing CDNs. This is also highlighted in the BIER WG use case draft [I-D.ietf-bier-use-cases], specifically with potential benefits in terms of utilizing multicast in the delivery of content but also reducing load on origin as well as delegation server. We return to this benefit in the trial experiences in Section 5.
Edge networks often see the deployment of novel network level technology, e.g., in the space of IoT. Such IoT deployments have for many years relied, and often still do, on proprietary protocols for reasons such as increased efficiency, lack of standardization incentives and others. Utilizing the underlay deployment configuration in Section 3.3.2, application gateways/proxies can integrate such edge deployments into IP-based services, e.g., utilizing CoAP [RFC7252] based machine-to-machine (M2M) platforms such as oneM2M [oneM2M] or others.
Another area of increased edge network innovation is that of mobile (access) networks, particularly in the context of the 5th generation of mobile networks (often called "5G"). With the proliferation of network softwarization (using technologies like service orchestration frameworks leveraging NFV and SDN concepts) access networks and other network segments, the ICN-as-a-Slice deployment configuration in Section 3.4 provides a suitable migration path for integration non-IP-based edge networks into the overall system through virtue of realizing the relevant (ICN) protocols in an access network slice.
Migrating the core network of the Internet requires not only significant infrastructure renewal but also the fulfillment of the significant performance requirements, particularly in terms of throughput. For those parts of the core network that would see a migration to an SDN-based optical transport, such as proposed by major operators such as AT&T, the ICN-as-a-Slice deployment configuration in Section 3.4 could see the introduction of native ICN solutions within slices provided by the SDN-enabled transport network or as virtual network functions, allowing for isolating the ICN traffic while addressing the specific ICN performance benefits and constraints within such isolated slice. For ICN solutions that natively work on top of SDN, the underlay deployment configuration in Section 3.3.1 provides an additional migration path, preserving the IP-based services and applications at the edge of the network, while realizing the core network routing through an ICN solution (possibly itself realized in a slice of the SDN transport network).
In this section, we will outline trial experiences, often conducted within international collaborative project efforts. Our focus here is on the realization of the various deployment configurations in Section 3, and we therefore categorize the trial experiences according to these deployment configurations. While a large body of work exists at the simulation or emulation level, we specifically exclude these studies from our presentation to retain the focus on real life experiences.
Although the FP7 PURSUIT [IEEE_Communications] efforts were generally positioned as a wholesale replacement of IP (Section 3.1), the project realized its experimental test bed as an L2 VPN-based overlay between several European, US as well as Asian sites, i.e., following the overlay deployment configuration presented in Section 3.2. Software-based forwarders were utilized for the ICN message exchange, while native ICN applications, e.g., for video transmissions, were showcased. At the height of the project efforts, about 70+ nodes were active in the (overlay) network with presentations given at several conferences as well as to the ICNRG.
The Network of Information (NetInf) is the approach to Information-Centric Networking developed by the European Union (EU) FP7 SAIL project (http://www.sail-project.eu/). NetInf provides both name-based forwarding with CCNx-like semantics and name resolution (for indirection and late-binding). The NetInf architecture supports different deployment options through its convergence layer abstraction. In its first prototypes and trials, NetInf was deployed mostly in an HTTP embedding and in a UDP overlay following the overlay deployment configuration in Section 3.2. Reference [SAIL_NetInf] describes several trials including a stadium environment large crowd scenario and a multi-site testbed, leveraging NetInf’s Routing Hint approach for routing scalability.
The Named Data Networking (NDN) is one of the research projects funded by the National Science Foundation (NSF) of the USA as part of the Future Internet Architecture Program. The original NDN proposal was positioned as a wholesale replacement of IP (Section 3.1). However, in several trials, NDN generally follows the overlay deployment configuration of Section 3.2 to connect institutions over the public Internet across several continents. The use cases covered in the trials include real-time video-conferencing, geo-locating, and interfacing to consumer applications. Typical trials involve several hundred NDN enabled nodes (https://named-data.net/ndn-testbed/).
ICN2020 is an ICN related research project funded by the EU as part of the H2020 research and innovation program (http://www.icn2020.org/). ICN2020 has a specific focus to advance ICN towards real-world deployments through innovative applications and global scale experimentation. Both NDN and CCN approaches are within the scope of the project.
ICN2020 was kicked off in late 2016 and so has not yet published results relating to deployment trials. The plan, however, is to involve ICN testbeds in EU, Japan and the USA and federate them. The GEANT testbed (https://www.geant.org/) is being considered as one means to federate the different ICN testbeds in the overlay deployment configuration of Section 3.2 over the public Internet.
POINT and RIFE are two more ICN related research projects funded by the EU as part of the H2020 effort. The efforts in the H2020 POINT+RIFE projects follow the underlay deployment configuration in Section 3.3.1, although this is mixed with utilizing an overlay deployment to provide multi-national connectivity. However, underlay SDN-based deployments do exist at various project partner sites, e.g., at Essex University, without any overlaying being realized. Edge-based network attachment points (NAPs) provide the IP/HTTP-level protocol mapping onto ICN protocol exchanges, while the SDN underlay (or the VPN-based L2 underlay) is used as a transport network.
The multicast as well as service endpoint surrogate benefits in HTTP-based scenarios, such as for HTTP-level streaming video delivery, have been demonstrated in the deployed POINT test bed with 80+ nodes being utilized. Demonstrations of this capability have been given to the ICNRG in 2016, while public demonstrations were also provided at events such as Mobile World Congress in 2016 [MWC_Demo]. The trial has also been accepted by the ETSI MEC group as a proof-of-concept with a demonstration at the ETSI MEC World Congress in 2016.
While the afore-mentioned demonstrations all use the overlay international deployment, both H2020 efforts plan ICN underlay trials in the summer and fall of 2017. One such trial will involve commercial end users located in the Primetel network in Cyprus with the use case centered on IPTV and HLS video dissemination. Another trial is planned for fall 2017 in the community network of "guifi.net" in the Barcelona region, where the solution will be deployed in 40 households, providing general Internet connectivity to the residents. Standard IPTV STBs as well as HLS video players will be utilized in accordance with the aim of this deployment configuration, namely to provide application and service migration.
Starting in January 2017, the H2020 FLAME efforts aims at providing an experimental ground for the aforementioned POINT/RIFE solution in initially two city-scale locations, namely in Bristol and Barcelona. This trial will again follow the underlay deployment configuration in Section 3.3.1 as per POINT/RIFE approach. Currently, experiments are ongoing,conducted by the city/university joint venture Bristol-is-Open (BIO), to ensure the readiness of the city-scale SDN transport network for such experiments. A third trial of the aforementioned ETSI MEC PoC is planned for mid 2017. This trial will showcase operational benefits provided by the ICN underlay for the scenario of a location-based game. These benefits aim at reduced network utilization through improved video delivery performance (multicast of all captured videos to the service surrogates deployed in the city at six locations) as well as reduced latency through the playout of the video originating from the local NAP instead of a remote server.
Ensuring the technology readiness and the early trialing of the ICN capabilities lays the ground for the goal of the H2020 FLAME efforts to conduct 23 large-scale experiments in the area of Future Media Internet (FMI) throughout 2018 and 2019. Standard media service functions as well as applications will ultimately utilize the ICN underlay in the delivery of their experience. The platform, which includes the ICN capabilities, will utilize concepts of SFC, integrated with NFV and SDN capabilities of the infrastructure. The ultimate goal of these platform efforts is the full integration of ICN into the overall media function platform for the provisioning of advanced (media-centric) internet services.
The work in [White] proposes an underlay deployment configuration based on Section 3.3.1. The use case is ICN for content distribution within CDN server farms (which can be quite large and complex) to leverage ICN's superior in-network caching properties. This "island of ICN" based CDN is then used to service standard HTTP/IP-based content retrieval request coming from the general Internet. This approach acknowledges that whole scale replacement (see Section 3.1) of existing HTTP/IP end user applications and related Web infrastructure is a difficult proposition. [White] does not yet provide results but indicated that experiments will be forthcoming.
The ICN Research Challenges [RFC7927] describes key ICN principles and technical research topics. As the title suggests, [RFC7927] is research oriented without a specific focus on deployment or standardization issues. This section addresses this open area by identifying key protocol functionality that that may be relevant for further standardization effort in IETF. The focus is specifically on identifying protocols that will facilitate future interoperable ICN deployments correlating to the scenarios identified in the deployment migration paths in Section 4. The identified list of potential protocol functionality is not exhaustive.
End user applications and services need a standardized approach to trigger ICN transactions. For example, in Internet and Web applications today, there are established socket APIs, communication paradigms such as REST, common libraries, and best practices. We see a need to study application requirements in an ICN environment further and, at the same time, develop new APIs and best practices that can take advantage of ICN communication characteristics.
A key issue in CDNs is to quickly find a location of a copy of the object requested by an end user. In ICN, a Named Data Object (NDO) is typically defined by its name. There already exists [RFC6920] that is suitable for static naming of ICN data objects. Other ways of encoding and representing ICN names have been described in [I-D.irtf-icnrg-ccnxmessages] and [I-D.mosko-icnrg-ccnxurischeme]. Naming dynamically generated data requires different approaches (for example, hash digest based names would normally not work), and there is lack of established conventions and standards.
Another CDN issue for ICN is related to multicast distribution of content. Existing CDNs have started using multicast mechanisms for certain cases such as for broadcast streaming TV. However, as discussed in Section 5.2.1, certain ICN approaches provide substantial improvements over IP multicast, such as the implicit support for multicast retrieval of content in all ICN flavours.
Caching is an implicit feature in many ICN architectures that can improve performance and availability in several scenarios. The ICN in-network caching can augment managed CDN and improve its performance. The details of the interplay between ICN caching and managed CDN need further consideration.
ICN provides the potential to redesign current edge and core network computing approaches. Leveraging ICN’s inherent security and its ability to make name data and dynamic computation results available independent of location, can enable a secure, yet light-weight insertion of traffic into the network without relying on redirection of DNS requests. For this, proxies that translate from commonly used protocols in the general Internet to ICN message exchanges in the ICN domain could be used for the migration of application and services within deployments at the network edge but also in core networks. This is similar to existing approaches for IoT scenarios where a proxy translates CoAP request/responses to other message formats. For example, [RFC8075] specifies proxy mapping between CoAP and HTTP protocols. However, as mentioned previously, ICN will allow us to evolve the role of gateways/proxies as ICN message security should be preserved through the protocol translation function of a thus offer a substantial gain. Another area is integration of ICN into networks that support virtualized infrastructure in the form of NFV/SDN. Further work is required to validate this idea and document best practices.
Without claiming completeness, Table 1 maps the open the open ICN issues identified in this document to potential protocol efforts that could address some aspects of the gap.
| ICN Gap | Potential Protocol Effort |
|---|---|
| 1-Support of | HTTP/CoAP support of ICN semantics |
| REST APIs | |
| 2-Naming | Dynamic naming of ICN data objects |
| 3-Multicast | Multicast enhancements for ICN |
| distribution | |
| 4-In-network | ICN Cache placement and sharing |
| caching | |
| 5-NFV/SDN | Integration of ICN with NFV/SDN |
| Support | |
| 6-ICN | Mapping of HTTP and other protocols onto ICN |
| mapping | message exchanges (and vice-versa) while preserving ICN message security |
This document provides high level deployment considerations for the ICN community. Specifically, the major configurations of possible ICN deployments are identified as (1) wholesale replacement of existing Internet infrastructure; (2) ICN-as-an-Overlay; (3) ICN-as-an-Underlay; and (4) ICN-as-a-Slice. Existing ICN trial systems mainly fall either under the ICN-as-an-Overlay or ICN-as-an-Underlay configuration.
In terms of deployment migration paths, ICN-as-an-Underlay offers a clear migration path for existing CDN, edge and core networks to go to an ICN paradigm. ICN-as-a-Slice is an attractive deployment option for future 5G systems (i.e., for 5G radio and core networks) which will naturally support network slicing, but this still has to be validated through actual trial experiences. Finally, for the crucial issue of existing application and service migration to ICN, various mapping schemes are possible to mitigate impacts. For example, HTTP/TCP/IP flows may be mapped to ICN message flows at a proxy in the ICN-as-an-Underlay configurations leaving the massive number of existing end point applications/services untouched or minimally impacted.
Finally, this document describes a set of technical features in ICN that warrant potential future IETF specification work. This will aid initial and incremental deployments to proceed in an interoperable manner. The fundamental details of the potential protocol specification effort, however, are best left for future study by the appropriate WGs and/or BoFs.
This document requests no IANA actions.
ICN was purposefully designed from the start to have certain intrinsic security properties. The most well known of which are authentication of delivered content and (optional) encryption of the content. [RFC7945] has an extensive discussion of various aspects of ICN security including many which are relevant to deployments. Specifically, [RFC7945] points out that ICN access control, privacy, security of in-network caches, and protection against various network attacks (e.g. DoS) have not yet been fully developed due to the lack of real deployments. [RFC7945] also points out relevant advances occurring in the ICN research community that hold promise to address each of the identified security gaps. Lastly, [RFC7945] points out that as secure communications in the existing Internet (e.g. HTTPS) becomes the norm, that major gaps in ICN security will inevitably slow down the adoption of ICN.
In addition to the security findings of [RFC7945], this document has highlighted that all anticipated ICN deployment configurations will involve co-existence with existing Internet infrastructure and applications. Thus even the basic authentication and encryption properties of ICN content will need to account for interworking with non-ICN content to preserve end-to-end security. For example, in the edge network underlay deployment configuration described in Section 3.3.2, the gateway/proxy that translates HTTP or CoAP request/responses into ICN message exchanges will need to support a model to preserve end-to-end security.
The authors want to thank Alex Afanasyev, Xavier de Foy, Hannu Flinck, Dave Oran, and Prakash Suthar for their very useful reviews and comments to the document.