Towards Edge Computing as a Service: Dynamic Formation of the Micro Data-Centers
0 likes149 views
The document proposes a model for dynamically organizing edge computing nodes into micro clouds to provide edge computing as a service. The model involves grouping nodes into clusters, clusters into regions, and regions into a topology. Micro clouds are ephemeral cloud-like structures serving local requests before reaching the traditional cloud. Protocols for health checking, cluster formation, and listing the system state are proposed. The model is inspired by cloud architecture and aims to lower latency by processing data closer to its source.
Towards Edge Computing as a Service: Dynamic Formation of the Micro Data-Centers
- 1. Towards edge computing as a service: dynamic formation of the micro data-centers Miloš Simić Ivan Prokić Jovana Dedeić Goran Sladić Branko Milosavljević Univerzitet u Novom Sadu Serbia
- 2. Outline Introduction The model Formal specifications Conclusions Towards edge computing as a service: our proposal The dynamic organization of geo-distributed edge nodes into micro clouds Our model: I edge computing nodes are organized into micro clouds dynamically, abstracting infrastructure to the level of software — infrastructure as software I is inspired by the cloud architecture, with adaptations for a different environment I geo-distribution means in proximity to some large populations, where micro clouds serve user requests locally first A synergy of edge and cloud computing 1 / 15
- 3. Outline Introduction The model Formal specifications Conclusions The problem I Edge computing is a paradigm where computing happens close to the data source, to lower the latency for its clients by contacting the cloud only when needed 2 / 15
- 4. Outline Introduction The model Formal specifications Conclusions Related work Platform models: I Kubernetes I Nebula I OpenStack Nodes organization: I Zone-based organization I Micro data centers - serve nearby population (theory) I Nano data centers - serve single household (using SDN) I Drop computing - ad hoc formation 3 / 15
- 5. Outline Introduction The model Formal specifications Conclusions Organization of the nodes Organization: I group of nodes form a cluster I group of clusters form a region I group of regions form a topology Edge centric computing Cloud computing Topology (logical) Cloud provider (logical) Region (logical) Region (physical) Cluster (physical) Zone (physical) Table: Similar concepts between cloud and edge-centric computing 4 / 15
- 6. Outline Introduction The model Formal specifications Conclusions Edge-centric computing as a service architecture with separation of concerns Cloud Cluster 1 Cluster n Region 1 *** Topology 1 Cluster 1 Cluster n Region n *** Cluster 1 Cluster n Region 1 Topology n Cluster 1 Cluster n Region n *** *** Core ECC Clients Services Devices Resources *** *** *** *** 5 / 15
- 7. Outline Introduction The model Formal specifications Conclusions Micro clouds I ephemeral cloud-like structures serving local requests first, before reaching to the traditional cloud I their size and existence are defined by the local population using them I clients have illusion that communicate with the traditional clouds I designed for failure using automated tools where no micro cloud is irreplaceable 6 / 15
- 8. Outline Introduction The model Formal specifications Conclusions Physical capabilities I capabilities of ARM-based devices have a good performance for building servers and clusters, considering their performance per Watt relation (Aroca et al., J. Parallel Distributed Comput., 2012) I these servers can be spread in base stations, coffee shops, or over geographic regions to avoid latency, and huge bandwidth (Wang et al., IEEE Access, 2017, Monsalve et al., Future Gener. Comput. Syst., 2018) I they can serve as firewalls and pre-processing tier for the cloud (Satyanarayananet al., EDGE 2019) I users get a unique ability to dynamically and selectively control the information sent to the cloud (Satyanarayananet al., EDGE 2019) 7 / 15
- 9. Outline Introduction The model Formal specifications Conclusions Formation of micro clouds and the protocols The system we propose uses remote configuration and it relies on three protocols: I health-check protocol informs the system about the state of every node I cluster formation protocol forms new clusters I list detail protocol shows the current state of the system to the user 8 / 15
- 10. Outline Introduction The model Formal specifications Conclusions Proof of concept implementation Gateway Queue State service events Nodes service events Command push service rpc Authentication & Authorization Service rpc rpc rpc rpc rpc rpc rpc Log service log events rpc http/s log events CLI 9 / 15
- 11. Outline Introduction The model Formal specifications Conclusions Possible real-life implementations We see two possible scenarios for implementation: I a stand-alone implementation I integration within existing tools, as a node organizer and register Application areas: I there are some ideas in the paper 10 / 15
- 12. Outline Introduction The model Formal specifications Conclusions Formal analysis? If there is a bug in a distributed algorithm, no matter how improbable it may seem, it’s not a question of whether it will appear, it’s a question of when it will appear. (Leslie Lamport, Heidelberg Laureate Forum 2021, https://www.youtube.com/watch?v=KVs3YFKqclU) New York is a city in which one in a million thing happen to eight people a day. (Whitfield Diffie, Heidelberg Laureate Forum 2021) 11 / 15
- 13. Outline Introduction The model Formal specifications Conclusions Formally specifying our protocols We were searching for: I a formalism that is proven correct I and expressive enough I but also easy to follow Our first candidate was multiparty asynchronous session types (K. Honda, N. Yoshida, and M. Carbone, Multiparty asynchronous session types, POPL 2008) 12 / 15
- 14. Outline Introduction The model Formal specifications Conclusions Formally specifying our protocols We were searching for: I a formalism that is proven correct I and expressive enough I but also easy to follow Our first candidate was multiparty asynchronous session types (K. Honda, N. Yoshida, and M. Carbone, Multiparty asynchronous session types, POPL 2008) but it was too rigorous... So we asked Nobuko for a help and she pointed us to Explicit connection actions in multiparty session types (R. Hu and N. Yoshida, Explicit connection actions in multiparty session types, FASE 2017) 12 / 15
- 15. Outline Introduction The model Formal specifications Conclusions Multiparty asynchronous session types Work of Hu and Yoshida follows the top-down approach of (which was quite easy to follow) I first define global type I then obtain local types through projections Syntax of global types: G ::= {p † qi:`i(Ti).Gi}i∈I | µt.G | t | end where † ∈ {→, } and I 6= ∅. Syntax of local types: S ::= +{qiα`i(Ti).Si}i∈I | µt.S | t | end where α ∈ {!, !!} or α ∈ {?, ??} (in which case qi = qj must hold for all i, j ∈ I, to ensure consistent external choice subjects), and I 6= ∅. 13 / 15
- 16. Outline Introduction The model Formal specifications Conclusions Health-check protocol State Nodes Node healthcheck opt alive used active update Log free used free G1 = nodenodes:health check(T1).
- 17. nodesstate:active(T2).nodeslog:used(T2) nodeslog:free(T2) Snode = nodes!!health check(T1) Snodes = node??health check(T1).+
- 19. Outline Introduction The model Formal specifications Conclusions Concluding remarks The dynamic organization of geo-distributed edge nodes into micro clouds Our model: I edge computing nodes are organized into micro clouds dynamically I is inspired by the cloud architecture Thank you for your attention! (M. Simić, I. Prokić, J. Dedeić, G. Sladić and B. Milosavljević, ”Towards Edge Computing as a Service: Dynamic Formation of the Micro Data-Centers,” in IEEE Access, vol. 9, pp. 114468-114484, 2021, doi: 10.1109/ACCESS.2021.3104475.) 15 / 15