# Notes taken during context of the discussions

Topics/Keywords:

Topic 1:
Privacy/Security
Digital Sovereignty
Adversarial Benchmarking
Tamper-resistant processing

Topic 2:
Edge Computing
Computational Storage
In-Network Computing
Resource Disaggregation
Use cases, Applications, Workloads, Testbeds

Manynets
Transport Protocols
Measurements, Reproducibility
COVID-19 Impact

===================================================================================================


Edge Computing
Edge definitions other than the ones that Nitinder showed?
- Edge on premise stuff of companies
- (Fog)

Which resources in which places? Who would be running those? How diverse could those be?
Administrative/security domains?

Resource management? Different quantities of resources, constraints, boundaries?

Data generation? Processing on the edge? Security implications?
Where does edge happen and who makes it happen?
- private clouds (e.g. Oracle)
- deployment vs management

Is your edge stable?
- mobile ISPs
- predictability; switching between mutlptiple providers, trust implications

edge as distribution of administration
- one entity (less complex, more centralized) vs multiple ones (more complex, less centralized)
- different domain: data and application?

where does edge stop and where does cloud begin?
- different views, top-down vs bottom-up
- allocation/mapping of resources and users
- end-user device vs third-party entity
    - depends on who controls and manages it?

most important aspects of Edge:
- security/privacy? (smart home or IoT devices)
- latency? (interactive applications)

trust relationships might differ depending on edge infrastructure
(heterogeneity of providers, laws/regulations,
what is the companies main business (data selling vs something else)?)

different stakeholders: ISP as edge provider, rather than only as transit points
- better or worse privacy? handing data to other entity than large cloud providers?

-> how much of the traffic would still exist (or where does it go to)
if we have edge instead of clouds?
- depends on customers and their uses

Shifts in Internet traffic due to lockdowns:
- diurnal patterns, now consistent throughout the day
- more video streaming
- elasticity of services, capacity does not have to be ramped up immediately
    (some downscaling of bitrate, e.g., Netflix or YouTube)
- networks might be overprovisioned


Moving content to the edge? main point of edge rather processing
-> for processing purposes, synchronization between multiple parties makes it difficult
- consistency of data, frequency of updates to be considered
- read-only data might be preferred over data that requires many writes

5G deployments: performance not as strict as promised by providers
- what comes first: edge boxes vs wireless deployments

cost of edge storage/processing power/usage/maintenance?

remote operations
- video processing closeby on the edge rather than sending it to the cloud

applications: off-loading from cloud
- allocation of tasks + data to correct edge nodes
- remote attestation

business model: who pays for deployment of all the nodes and units?
- profit of company: paid directly (money) or paid by data?
- economics: edge might widen the wealth gap
- incremental deployments, natural growth of business
- can new deployments bring more money (economic feasibility)?
    - where would the additional money be invested into?
- initial customers/demand

more profitable:
- optimization/fixing of current technologies vs deployment of edge nodes?

service subscriptions over purchases
- dependency on other services (connectivity to network/Internet)
- special subscriptions: premium for better service (lower latency, better performance, ...)

===================================================================================================

Digital Sovereignty
- open technologies; getting rid of technological dependencies from other countries
- build up new Internet?
    maybe not necessary (China, Russia)

- definition: state-wide vs individual

- construct IT system (maybe with slightly worse performance)
    but with other more valuable assets, which would make it worth it

- diversification and distribution, where trust is not major weakness
    replacability of components on-the-fly
    combination of different solutions (due to them having specific fundametal weaknesses)
    cross-layer
    interoperability for independence of single technologies

is it required to multiply current resources for that purpose?
    abstractions for current variety of technologies and instructions
-> performance implications?

critical technologies or expertise/know-how (which we currently do not have access to)

ensure code safety
- formal verification
- measurements
- containment approaches
- multiple approaches required

does the general public want digital sovereignty?
    maybe ask who care about the next decade instead
    experts are aware: big problem, however, cannot be solved immediately

our role as scientists: not many resources required but instead clever approach
    also with better time to market
bring up solution that has immediate/initial use cases, rather than multi-stage

areas in which sovereignty needs to be pushed for
    important/critical infrastrucutre (plant, car, manifacturing site, plane, administration)

share of public funding, then prove the worth of it
European commission agenda 2019-2024
https://ec.europa.eu/commission/sites/beta-political/files/political-guidelines-next-commission_en.pdf

not only hardware but also data

importance of QoS?

other national technologies/companies (e.g., Lufthansa),
whereas for computational things primarily foreign solutions

governmental projects (Corona Warn-App, Siemens, ...) vs consumer uses

cost vs worth
probability too expensive to have sovereignty in all layers and areas of life

"lock-in" to specific contracts and service providers, migration difficult/impossible

important topic to work on due to large potential impact, good opportunity
(also encourages collaboration between different departments of TUM)