Networked Systems H (2021-2022)
Lecture 10: Future Directions
The final lecture will briefly review the course aims and contents, then conclude with a forward-looking discussion of possible future directions in which the network might evolve.
Part 1: Future Directions
The final part of the lecture reviews how the Internet is changing, to reduce latency, improve security, and avoid protocol ossification, and discusses some of longer-term research work driving the evolution of the network.
00:00:00.466 In this final lecture, I want to
00:00:02.300 talk about some possible future directions for
00:00:04.666 the development of the network.
00:00:07.900 So this course has focused on how
00:00:12.200 the Internet can change and evolve to address
00:00:14.966 some coming challenges.
00:00:18.233 It’s focused on the issues of how we establish
00:00:21.200 connections in an increasingly fragmented network,
00:00:24.466 thinking about the issues with network address
00:00:28.266 translation, the issues with the rise of
00:00:32.900 IPv6 and dual-stack hosts,
00:00:35.633 and I’ve spoken in some detail about
00:00:38.700 the challenges in establishing connections when the
00:00:42.133 machines are not necessarily in a common
00:00:44.166 addressing realm, and when there are multiple
00:00:46.866 different ways of potentially reaching a machine.
00:00:49.666 And this is techniques such as
00:00:51.966 the ICE algorithm for NAT traversal, the
00:00:54.633 Happy Eyeballs technique for connection racing for IPv6.
00:01:00.366 I’ve spoken about some of the issues with encryption, and
00:01:04.733 protecting against pervasive network monitoring,
00:01:08.433 and protecting against, and preventing,
00:01:12.033 transport ossification.
00:01:13.966 And this led to some of the
00:01:15.966 design of QUIC, with the entire protocol,
00:01:19.333 including the overwhelming majority of the transport
00:01:22.566 layer headers, being encrypted, and those which
00:01:24.900 are not encrypted being greased in order
00:01:27.533 to allow evolution.
00:01:29.566 And that's partly a security measure,
00:01:31.866 and it's partly an evolvability and
00:01:33.533 anti-ossification measure.
00:01:35.333 It’s looking at ways in which
00:01:37.666 we can keep changing the protocols by
00:01:39.833 deploying encryption to prevent middleboxes
00:01:42.333 interfering with our communications.
00:01:46.033 And I’ve spent a fair amount of
00:01:47.366 time talking about how we can reduce
00:01:49.166 latency, and support real-time and interactive content.
00:01:53.800 And, partly, this comes in, again,
00:01:55.800 in the design of protocols like QUIC,
00:01:58.133 in the design of TLS 1.3 with
00:02:01.400 reducing the number of round trips needed
00:02:03.166 to set up a connection.
00:02:06.066 It comes in, in
00:02:08.300 systems like content distribution networks that move
00:02:11.300 content nearer the edges, near the customers,
00:02:14.300 to reduce latency.
00:02:16.633 And it comes in, in the design
00:02:18.166 of real-time applications and protocols like RTP.
00:02:22.766 And I spoke about some of the
00:02:24.166 issues with congestion control, wireless networks,
00:02:27.366 and content distribution, and congestion control,
00:02:30.866 and how to make much more adaptive applications.
00:02:38.000 We’ve considered some of the challenges in
00:02:41.466 identifying content,
00:02:43.766 content distribution networks and naming, and how
00:02:48.600 you can securely
00:02:50.833 find the names for a piece of
00:02:53.233 content on the Internet that you want
00:02:54.866 to access, and how to do that
00:02:56.633 without being subject to phishing attacks.
00:02:59.366 And some of the challenges, and the
00:03:02.300 tussle for control of the DNS and naming,
00:03:05.666 and how those relate to censorship and
00:03:08.300 filtering, but also how the DNS can
00:03:11.200 be used to support content distribution networks.
00:03:15.100 And we've spoken about routing, and efficient
00:03:19.066 content delivery, in the last lecture.
00:03:22.766 And some of this leads into discussions
00:03:25.233 about decentralisation of the network, and the
00:03:28.533 rise of hyper-giants and content distribution networks,
00:03:31.866 that centralise content onto
00:03:35.533 a small number of providers.
00:03:41.000 As we've seen, there’s a large number of challenges.
00:03:45.700 And as a result of that,
00:03:47.000 the Internet is actually in the middle
00:03:48.700 of one of its most significant periods
00:03:50.933 of change, that certainly I’ve seen in
00:03:54.933 the time I’ve been involved in networking.
00:03:58.333 We're seeing IPv6 beginning to be significantly deployed.
00:04:04.600 And, partly, this is providing for increased address space,
00:04:09.700 and increasing numbers of devices on the network.
00:04:12.900 But it is also flexible enough,
00:04:15.533 because of the size of that address
00:04:17.300 space, that people are starting to look
00:04:19.766 at what they can do with IPv6
00:04:22.200 to evolve the way the network is being developed.
00:04:26.266 It's flexible, in that it’s got enough
00:04:28.733 bits in the address, that semantics can
00:04:31.733 be assigned to the addresses. So bits
00:04:34.433 can have meaning other than, perhaps,
00:04:36.233 just the location of a device on the network.
00:04:39.400 And it's got a very flexible header
00:04:41.533 extension mechanism that can be used to
00:04:44.400 layer-in features,
00:04:46.733 provide extra semantics,
00:04:51.433 and provide application semantics, as part of
00:04:53.966 the packet headers, to allow special processing.
00:04:56.500 And people are starting to explore the
00:04:58.566 things you can do with IPv6 as
00:05:00.766 it gets more widely deployed.
00:05:04.733 We've seen TLS 1.3 be rolled-out,
00:05:07.966 and massively improve and simplify security.
00:05:11.566 And we've seen it be incorporated into
00:05:13.633 the QUIC protocol, as the basis for
00:05:16.200 future transport evolution.
00:05:19.233 And, I've described QUIC as essentially a
00:05:22.133 better version of TCP, or as an
00:05:25.000 encrypted version of TCP, which combines the
00:05:28.100 goals TCP and TLS, and also adds
00:05:31.166 this idea of multi-streaming.
00:05:33.533 But I think QUIC is actually going
00:05:35.133 to be the basis for a lot
00:05:37.000 more developments, and a lot more evolution.
00:05:40.366 We’re already seeing this, to some extent.
00:05:44.633 There is already a datagram extension to
00:05:47.533 QUIC going through the standards process,
00:05:50.233 to start supporting real-time applications effectively over
00:05:55.100 QUIC, and it's pretty clear that we're
00:05:57.233 going to see a lot more evolution
00:05:58.833 and development in that space, with people using QUIC
00:06:02.166 as the basis for future transport protocols,
00:06:05.100 for real-time and interactive applications, and so on.
00:06:10.733 And this has led to the coming,
00:06:13.700 I think, adoption of HTTP/3 as the
00:06:15.733 basis for evolving the web,
00:06:18.900 and HTTP growing beyond web documents to
00:06:21.966 include a much richer set of real-time
00:06:25.733 and interactive services.
00:06:30.866 And, in parallel to this, I think
00:06:32.733 we've seen the rise of changes to
00:06:35.100 the DNS, many ways of running DNS
00:06:38.266 over encryption, whether it's a DNS over
00:06:40.666 HTTPS, or over TLS, or over QUIC,
00:06:44.833 in order to get secure name resolution.
00:06:49.300 And to avoid some of the control
00:06:51.200 points. And we’re seeing CDNs and overlays
00:06:55.033 increasingly making use of DNS
00:06:58.100 for directing hosts of the content.
00:07:04.466 And I think we're seeing an increasing
00:07:06.900 tussle for control, between the different industries
00:07:11.300 and the different providers.
00:07:14.133 On the one hand we've got
00:07:17.100 the model I’m describing, with QUIC,
00:07:20.100 and TLS, and HTTP/3, and encrypted DNS
00:07:25.066 to allow the application providers, and their
00:07:28.133 customers, and the end-users, to talk directly,
00:07:33.133 and to limit the visibility of the
00:07:35.033 network into that communication.
00:07:38.200 And, on the other hand, we have
00:07:40.066 operators trying to build application awareness into
00:07:43.300 their networks, trying to increase the communication
00:07:47.533 between the network and the endpoints,
00:07:49.433 to improve performance, and to sell enhanced services.
00:07:53.800 And there’s a tussle, where it’s not
00:07:55.266 clear how it's going to play out.
00:08:01.566 So that's the current set of developments
00:08:06.300 in the network. And that's the areas
00:08:08.433 where I've been trying to focus on
00:08:10.333 in this course, describing how the network
00:08:12.666 is currently changing.
00:08:16.400 In this last part, I’d like to
00:08:17.866 talk a little bit about some of
00:08:19.166 the longer-term challenges, some of the longer
00:08:21.533 term directions for the network, and think
00:08:24.166 about where the network might be going,
00:08:26.566 not in the next five years but
00:08:29.300 in the next 10 to 20 years.
00:08:35.633 And what might be the long-term future
00:08:37.900 developments of the Internet.
00:08:40.633 And, to be clear, what's coming in
00:08:43.233 the remainder of this part is speculative.
00:08:45.833 It’s my biased opinion of where I
00:08:48.766 see the network going, based on my
00:08:50.466 interests, based on the research that I
00:08:53.000 have seen happening.
00:08:55.433 But it's very much speculative. It may not come true.
00:09:02.133 But it's pointing to areas which I think
00:09:04.933 are interesting developments.
00:09:07.000 And nothing in this section is going to be assessed.
00:09:13.766 So where's the network going in the long term?
00:09:17.666 Well, I think, to get some understanding of
00:09:20.833 that, we need to look at the process by which
00:09:23.900 new ideas, new research, get incorporated into the network.
00:09:30.800 And, on the one hand, what we
00:09:32.466 see on the left of this slide, we have
00:09:34.266 the organisations that promote research
00:09:38.266 into computer networks.
00:09:40.166 The Association for Computing Machinery,
00:09:43.166 the USENIX Association, and the IEEE,
00:09:46.533 all of whom sponsor
00:09:50.100 both industrial and academic research in this
00:09:53.033 area, all of whom publish research in this area.
00:09:58.433 And this is the pure research side
00:10:01.666 of network development. This is people speculatively
00:10:05.533 trying to understand how the network could change.
00:10:08.900 In the middle, you have organisations like
00:10:12.166 the IRTF, the Internet Research Task Force,
00:10:15.500 which try to form the bridge between
00:10:19.200 these research organisations
00:10:20.666 and the standards organisations,
00:10:22.333 such as the IETF, which develop the
00:10:25.533 standards which we actually deploy.
00:10:29.000 And one of the other activities I have,
00:10:31.433 is that I chair the IRTF.
00:10:34.433 And the IRTF is a body which
00:10:38.100 promotes the evolution of the Internet.
00:10:40.433 It’s promoting the longer-term research and development
00:10:44.233 of the Internet protocols, and, as I say,
00:10:45.933 it's trying to bridge these organisations together.
00:10:49.400 And so by looking at some of
00:10:51.033 the work that's happening in the IRTF,
00:10:53.100 we can perhaps get an idea of
00:10:54.433 how the network might evolve, and what's
00:10:56.500 coming down the pipeline towards standardisation.
00:11:02.800 So the IRTF is organised as a
00:11:06.033 set of research groups, which focus on
00:11:08.366 longer-term development of ideas and protocols.
00:11:13.033 And it's organised to provide a forum where
00:11:15.633 the researchers and the engineers can
00:11:17.433 explore the feasibility of different research ideas.
00:11:20.766 And where the researchers,
00:11:23.833 developing ideas for the future of the
00:11:26.166 network, can learn from the engineers,
00:11:28.766 and the operators, who actually build and
00:11:30.500 operate the Internet.
00:11:32.166 But, equally, where the standards developers,
00:11:34.866 the engineers, the operations community, the implementors,
00:11:37.966 can learn from the research community.
00:11:40.566 Where the two can come together.
00:11:44.233 As I say, it’s organised as a
00:11:45.533 set of research groups. There’s currently 14
00:11:49.033 research groups that are listed on the
00:11:51.966 slide, I’ll talk about these in a
00:11:54.500 little more detail in a minute.
00:11:56.766 And as we can see, they’re covering a wide range of topics.
00:12:00.366 And there’s also an annual workshop we
00:12:03.433 organise, to help bring the communities together.
00:12:10.333 So what do the research groups do?
00:12:14.700 Well, they’re focused on several different topic areas.
00:12:19.766 One of which is the
00:12:22.866 space around security, and privacy, and human rights.
00:12:30.333 The cryptographic forum research group, the CFRG,
00:12:35.433 focuses on long-term development
00:12:38.566 of cryptographic primitives, and cryptographic techniques,
00:12:41.966 and guidance for using those techniques.
00:12:44.733 This is a research group looking at
00:12:47.700 new cryptographic algorithms, replacements for AES,
00:12:53.266 replacements for elliptic curve cryptography,
00:12:56.800 new elliptic curve algorithms, and the like.
00:13:01.700 And this is focused, very much, on
00:13:06.000 techniques, cryptographic techniques,
00:13:08.966 which support various privacy-enhancing
00:13:11.900 technologies. And it's beginning to focus on
00:13:15.166 post quantum cryptography, and cryptographic techniques that
00:13:19.566 can work in a world with working
00:13:22.500 quantum computers.
00:13:25.466 We have a privacy-enhancing technologies group,
00:13:29.633 which is focused on
00:13:32.933 the challenges of metadata in the network,
00:13:36.533 focusing on the challenges
00:13:39.833 of building a network that doesn't use
00:13:42.300 addresses, or that hides IP addresses,
00:13:45.266 in a way that prevents tracking.
00:13:47.466 And ways of providing privacy-enhancing logins,
00:13:53.166 and authentication tokens, and the like,
00:13:55.600 that can avoid tracking.
00:13:58.500 And we have a human rights protocol
00:14:00.966 considerations group, which is beginning to look
00:14:04.466 at, and understand, how Internet protocols and
00:14:07.066 standards affects human rights and privacy at
00:14:10.100 the Internet infrastructure level.
00:14:12.766 And it's looking at the
00:14:17.833 right of freedom of association on the
00:14:20.266 Internet, for example, and how that's affected
00:14:23.533 by protocol design.
00:14:25.866 It’s looking at
00:14:29.766 how protocols affect inclusivity and access,
00:14:34.733 and so on, and it's looking at
00:14:36.700 the politics of protocols.
00:14:39.866 And these three groups, are looking at
00:14:41.533 the interplay between security, privacy, and human rights,
00:14:44.300 and trying to raise awareness of
00:14:46.233 the broader societal and policy issues in
00:14:49.000 the standards community.
00:14:55.200 There's an interesting, I think, thread of
00:14:58.366 technical development, looking at the combination of
00:15:02.866 networks and distributed systems.
00:15:05.966 Looking at speculative new architectures for the internet,
00:15:09.700 which either emphasise data or emphasise computation.
00:15:15.366 If you think about the current network,
00:15:18.500 IP addresses identify devices, they identify attachment
00:15:22.233 points for devices in the network.
00:15:25.033 And these groups are looking at
00:15:28.066 the generalisation of content distribution networks,
00:15:31.566 and web caching infrastructure, and thinking about
00:15:34.266 what would happen if we replaced IP
00:15:36.366 addresses with content identifiers?
00:15:38.866 So the network would route towards particular
00:15:41.766 items of content, rather than routing packets
00:15:44.700 towards particular locations.
00:15:47.166 Or they’re looking at
00:15:50.100 generalising the network so it routes towards
00:15:54.300 addresses, and routes toward was named functions,
00:15:57.600 which are generalising the idea of serverless computation.
00:16:02.833 And the idea of both of these
00:16:04.000 groups is to think about what might
00:16:05.933 happen if you rearchitect the network around
00:16:08.333 either content, or computation, or both.
00:16:11.666 And think about the merger of communication,
00:16:16.066 data centres, computation, and data warehouses,
00:16:20.066 to form one large distributed system,
00:16:23.266 rather than an interconnection network which connects
00:16:26.933 compute devices, data stores, at the edges.
00:16:30.933 And thinking about what are the implications
00:16:34.000 for this change, towards a network with
00:16:36.333 ubiquitous data, or ubiquitous computation, for the
00:16:40.600 content provider/consumer relationship.
00:16:43.400 Thinking about will this help democratise the
00:16:46.500 network, will it help ensure a more
00:16:48.333 decentralised network.
00:16:50.966 will it help with hosting content throughout
00:16:54.233 the network in a way which empowers consumers,
00:16:56.866 or will it simply ossify the current
00:16:59.400 roles, and the current content distribution networks,
00:17:02.466 and large scale cloud providers.
00:17:04.400 And it’s looking at alternative architectures,
00:17:07.066 and how it can influence the way forward.
00:17:11.366 And all this leads to networks which
00:17:13.333 no longer have IP addresses as their
00:17:15.666 core, that no longer have the Internet
00:17:17.400 Protocol as their core, but are much
00:17:19.466 more about distributed computation and data.
00:17:27.000 There's a research group looking into a
00:17:29.833 technique, known as path aware networking.
00:17:33.866 And this is the idea of trying
00:17:35.566 to explore what can happen if we
00:17:38.166 make the applications, and the transport protocols,
00:17:41.500 much more aware of the network path,
00:17:44.300 and the characteristics of the network path.
00:17:47.233 Or, similarly, if we make the network
00:17:49.433 much more aware of the applications and
00:17:51.300 the transports that are running on it.
00:17:55.400 And this potentially has benefits, it potentially
00:17:58.133 has benefits for improving the quality of
00:18:00.800 service, for allowing applications to request special
00:18:04.800 handling in the network to improve performance,
00:18:08.700 and to maybe request low-latency service,
00:18:13.133 or specialised in network processing.
00:18:18.633 But, equally, it has potential challenges,
00:18:21.600 because it introduces a control point.
00:18:23.600 It introduces a way for the operators
00:18:26.333 to control the types of applications that
00:18:28.333 can run on the network.
00:18:30.866 And there are some significant questions around
00:18:33.066 trust, and privacy, and network neutrality,
00:18:36.266 which are relatively poorly understood.
00:18:40.600 And this is an area where we
00:18:41.933 see the IETF community currently seems determined
00:18:45.466 to enter a standardisation phase.
00:18:48.233 There's a technique called segment routing,
00:18:51.266 and segment routing in IPv6, SRv6,
00:18:54.733 which is starting to
00:18:57.500 work its way through the standardisation process,
00:19:00.300 and it's starting to get some traction,
00:19:02.433 which is building some of this application
00:19:04.933 awareness into the network infrastructure.
00:19:07.833 And there’s technique called APN,
00:19:10.266 which is an application-aware networking scheme,
00:19:12.600 that’s going in the same direction.
00:19:15.400 And, a number of
00:19:18.300 large Internet companies are pushing in the space.
00:19:24.733 In the IRTF, the research groups,
00:19:27.466 I think, they’re looking at some of
00:19:29.033 the more broad questions, trying to
00:19:31.333 understand what are the privacy implications,
00:19:33.466 what are the security implications, and what
00:19:37.200 are the incentives for both the endpoints to deploy
00:19:40.666 these features, for the applications to deploy
00:19:43.333 these path-aware features, and for the operators
00:19:46.266 to enable them. And how does it
00:19:50.100 shift the balance of control between the
00:19:52.166 applications, and the end-users, and the network operators.
00:19:56.633 And I think there's some interesting unsolved
00:19:58.466 questions in that space.
00:20:04.100 In the longer term, we have a
00:20:06.166 group looking at designing the quantum Internet.
00:20:11.233 And the idea, here, is that it
00:20:14.366 seems likely that people will
00:20:17.400 manage to build working, large-scale, quantum computers
00:20:22.166 in the next few years.
00:20:25.933 And if they do that, they will
00:20:27.466 want to network and interconnect those computers.
00:20:30.600 The quantum Internet group is looking at
00:20:34.700 how we can architect a network that
00:20:36.766 provides quantum entanglement as a service.
00:20:40.133 It’s looking at how to build global-scale
00:20:43.700 distributed quantum computers.
00:20:47.700 And this is very much the exchange
00:20:50.533 of Bell Pairs; it’s the exchange of
00:20:52.800 quantum entangled state.
00:20:57.166 And it’s leading to a surprisingly traditional
00:21:01.566 network architecture. A control plane that looks
00:21:05.100 like the control plane used in a
00:21:07.733 lot of Internet service provider networks for
00:21:10.433 traffic engineering.
00:21:11.766 But rather than managing circuits and traffic
00:21:14.800 flows, it manages the setup of optically
00:21:18.166 clear paths, which can be used to
00:21:20.933 transmit entangled photons,
00:21:23.300 to manage entangled quantum state.
00:21:27.266 And this group’s coming to the conclusion
00:21:31.133 of its architecture development phase, and is
00:21:34.300 starting to build experiments, starting to prototype
00:21:38.166 these systems, and see if they actually work.
00:21:41.566 And people are actually starting to build
00:21:43.800 the initial versions of the quantum Internet,
00:21:47.600 and do at least small-scale experiments with
00:21:50.500 networked quantum computers and quantum entanglement.
00:21:59.300 And, perhaps more pressingly,
00:22:02.533 we have a group, the Global Access
00:22:04.766 to the Internet for All group,
00:22:06.366 which is looking at global access and
00:22:07.966 sustainability. And it's looking about how to
00:22:10.466 address the global digital divide.
00:22:14.500 It's trying to share experiences and best
00:22:18.733 practices, foster collaboration in helping build,
00:22:22.866 and develop, and make effective use of
00:22:25.200 the Internet in rural, and remote,
00:22:27.733 and under-developed regions. And there’s a lot
00:22:30.966 of interest, a push towards community run,
00:22:34.833 community led, networks
00:22:38.133 to provide a more sustainable, more locally
00:22:42.100 run network, which reflects the needs of
00:22:45.733 the local communities, rather than the mega-corporations.
00:22:49.900 And it's trying to develop a shared
00:22:52.066 vision towards building a sustainable global network.
00:22:57.066 And, most of the focus here is
00:22:58.866 on developing countries, and on
00:23:01.866 building a fairer, more sustainable, network in
00:23:05.966 those parts of the world. But it's
00:23:08.266 also looking at access for less developed,
00:23:13.466 perhaps more rural regions, of
00:23:17.566 the world. And there's been some interesting
00:23:21.233 work trying to build community networks in
00:23:24.700 the Scottish Highlands and Islands, for example,
00:23:28.166 where there’s more constrained infrastructure.
00:23:32.033 But it's also talking about energy efficiency,
00:23:35.900 and renewable power, and building networks which
00:23:40.666 work much more sustainably.
00:23:46.066 And there are other groups, which I
00:23:48.466 don't have time to talk about in
00:23:49.866 detail, looking at measuring and understanding network
00:23:55.100 behaviour, in the measurement and analysis protocols group.
00:23:59.666 Looking at developing new congestion control,
00:24:04.666 network coding algorithms to improve performance and
00:24:08.133 make applications more adaptive.
00:24:10.733 Looking at intent- and
00:24:13.400 artificial intelligence-based approaches
00:24:15.833 to managing and operating networks.
00:24:19.200 Understanding the issues of trust, and identity
00:24:22.600 management, and name resolution, and resource ownership,
00:24:25.966 and discovery, in decentralised infrastructure networks.
00:24:29.966 And looking at some of the challenges,
00:24:32.800 the research challenges, from initial, broad,
00:24:35.833 real-world deployments of Internet-of-Things devices,
00:24:38.933 and how we can make those devices more sustainable,
00:24:42.066 more programmable, and more secure.
00:24:49.466 The key thing I want to get
00:24:52.100 across is that the network, the Internet, is not finished.
00:24:58.266 The protocols and fundamental design are still
00:25:01.400 evolving, they're still changing.
00:25:05.933 There's a,
00:25:08.200 perhaps, a view of networking you get
00:25:11.566 from reading various textbooks that
00:25:15.366 the Internet is IPv4, and TCP, and the web.
00:25:21.700 And it's always been that, and it always will be that.
00:25:25.833 But nothing could be further from the truth.
00:25:29.400 The fundamental infrastructure has
00:25:32.566 massively shifted over the last few years.
00:25:35.400 And I think we're in the middle of this enormous
00:25:37.433 transition, and we are getting rid of
00:25:40.133 IPv4, and we are getting rid of
00:25:42.833 TCP, and we're getting rid of HTTP/1.1
00:25:46.400 and the traditional web infrastructure.
00:25:49.066 With IPv6 and QUIC, we're seeing a
00:25:52.800 radical restructuring of both the network infrastructure
00:25:56.200 layer, the IP layer, to support more
00:25:59.566 addresses, and to support more programability,
00:26:02.433 to support more application semantics.
00:26:05.600 But also the transport and the web layers,
00:26:09.033 to replace TCP, and better support real-time
00:26:12.700 and multimedia transport, and to be more
00:26:14.533 secure and more evolvable.
00:26:17.100 And the network is in the middle
00:26:18.866 of this enormous shift.
00:26:22.333 And, looking forward, I think there are
00:26:23.933 potentially even more significant changes to come,
00:26:26.833 with a merger of computation and communication
00:26:31.233 and data centres as one
00:26:33.700 global-scale distributed system.
00:26:36.366 With some of the ideas around path awareness,
00:26:39.766 the quantum Internet,
00:26:40.800 some of the security and sustainability challenges.
00:26:44.366 The network is not finished.
00:26:46.233 The network is keeping changing.
00:26:48.266 There’s still some exciting developments to come.
00:26:56.100 And that’s, essentially, all I have for this course.
00:27:02.466 To wrap up.
00:27:05.633 There will, of course, be an assessment at the end.
00:27:10.933 There'll be a final exam, and it
00:27:14.566 will be worth the usual 80%,
00:27:17.666 and will be held in the April/May
00:27:19.200 time frame as expected.
00:27:23.800 The exam is structured as a set
00:27:26.066 of three questions. It’s an "answer all
00:27:29.500 three questions” rubric.
00:27:31.566 And it will be focused on testing
00:27:33.233 your understanding of networked systems.
00:27:38.033 When answering the exam questions, tell me
00:27:41.633 what you think, and justify your answers.
00:27:48.066 The type of online, open book,
00:27:51.366 exams that we are forced to do
00:27:54.866 these days, focus much more on deeper
00:27:59.200 understanding of material, and much less on
00:28:02.533 book-work and memorisation.
00:28:06.233 There's little point asking an exam question which
00:28:13.100 tests your memory, when you're doing this,
00:28:16.666 when you're doing the exam online from
00:28:19.633 home, and you have Google next to you.
00:28:22.966 So the questions will be focussed more
00:28:24.966 on testing your understanding, than on your
00:28:27.600 testing your recall.
00:28:30.466 There are past exam papers on moodle.
00:28:33.233 The past exam papers go back some number of years.
00:28:37.200 As you may perhaps expect, the exam
00:28:39.866 questions from 2020 are probably more representative
00:28:43.933 of the style of this year's exam
00:28:46.000 than the older papers, although there are
00:28:48.866 certainly questions in this style going back
00:28:51.933 for many years.
00:28:55.900 The assessed coursework, the marks will be
00:28:59.233 available shortly, and
00:29:02.733 I apologise that it's taken a little
00:29:05.333 while to talk to mark some of that.
00:29:08.333 There's no specific revision lecture, but we
00:29:11.566 have the Teams chat, and we have
00:29:13.200 email, so please get in touch if
00:29:15.033 you have questions about the material.
00:29:19.066 And, looking forward to next year,
00:29:21.566 if you're interested in doing Level 4
00:29:24.933 or MSci projects relating to networked systems,
00:29:27.966 then please get in touch with me,
00:29:29.900 send me email on the address.
00:29:32.566 I’m always very keen to work with
00:29:34.900 motivated students to develop projects.
00:29:38.633 My particular interests, I think, are around
00:29:42.900 improving Internet Protocol standards and specifications,
00:29:47.666 and working with the IETF and IRTF
00:29:50.600 communities to improve the way we build standards.
00:29:54.600 They’re about improving transport protocols,
00:29:57.200 real-time applications, and QUIC.
00:30:01.133 They’re about building alternative networking APIs,
00:30:05.466 and thinking about how we can use
00:30:07.566 modern, high-level, languages like Rust to change
00:30:11.600 the way we program networks, make network
00:30:14.300 programming easier, more flexible, and higher performance.
00:30:17.600 And they’re about measuring and understanding the network.
00:30:21.433 So if you have any interest in
00:30:23.266 any of those topics, please come and talk to me.
00:30:27.966 I tend to try and set projects,
00:30:30.700 do research, which has a strong focus
00:30:33.300 on interaction with the research communities,
00:30:36.800 interaction with the IETF standards.
00:30:39.333 And I have a range of project ideas,
00:30:42.633 and projects can go in a
00:30:43.866 range of different ways, some of which
00:30:45.833 are very strongly technical, some of which
00:30:48.333 focus much more heavily on the standardisation process,
00:30:52.100 and the way in which standards,
00:30:56.266 and protocols, are developed, and are looking
00:30:59.033 at the social and political aspects of
00:31:01.666 the way the Internet is developing.
00:31:04.033 So, as I say, if you have
00:31:05.333 an interest in any of these topics,
00:31:08.433 please come talk to me.
00:31:13.166 And that's all we have.
00:31:15.200 That's what I want to say about networks.
00:31:17.666 Thank you for your attention over the past few weeks.
00:31:22.600 I hope you have found some of the material interesting,
00:31:25.733 and if you have questions or comments
00:31:28.000 or things you'd like to discuss further,
00:31:29.666 please do get in touch. Thank you.
Discussion will be open-ended and student driven, focussed around the course material in general and the suggested future directions.