---
_id: '19607'
abstract:
- lang: eng
  text: "Modern services consist of modular, interconnected\r\ncomponents, e.g., microservices
    forming a service mesh. To\r\ndynamically adjust to ever-changing service demands,
    service\r\ncomponents have to be instantiated on nodes across the network.\r\nIncoming
    flows requesting a service then need to be routed\r\nthrough the deployed instances
    while considering node and link\r\ncapacities. Ultimately, the goal is to maximize
    the successfully\r\nserved flows and Quality of Service (QoS) through online service\r\ncoordination.
    Current approaches for service coordination are\r\nusually centralized, assuming
    up-to-date global knowledge and\r\nmaking global decisions for all nodes in the
    network. Such global\r\nknowledge and centralized decisions are not realistic
    in practical\r\nlarge-scale networks.\r\n\r\nTo solve this problem, we propose
    two algorithms for fully\r\ndistributed service coordination. The proposed algorithms
    can be\r\nexecuted individually at each node in parallel and require only\r\nvery
    limited global knowledge. We compare and evaluate both\r\nalgorithms with a state-of-the-art
    centralized approach in extensive\r\nsimulations on a large-scale, real-world
    network topology.\r\nOur results indicate that the two algorithms can compete
    with\r\ncentralized approaches in terms of solution quality but require\r\nless
    global knowledge and are magnitudes faster (more than\r\n100x)."
author:
- first_name: Stefan Balthasar
  full_name: Schneider, Stefan Balthasar
  id: '35343'
  last_name: Schneider
  orcid: 0000-0001-8210-4011
- first_name: Lars Dietrich
  full_name: Klenner, Lars Dietrich
  last_name: Klenner
- first_name: Holger
  full_name: Karl, Holger
  id: '126'
  last_name: Karl
citation:
  ama: 'Schneider SB, Klenner LD, Karl H. Every Node for Itself: Fully Distributed
    Service Coordination. In: <i>IEEE International Conference on Network and Service
    Management (CNSM)</i>. IEEE; 2020.'
  apa: 'Schneider, S. B., Klenner, L. D., &#38; Karl, H. (2020). Every Node for Itself:
    Fully Distributed Service Coordination. In <i>IEEE International Conference on
    Network and Service Management (CNSM)</i>. IEEE.'
  bibtex: '@inproceedings{Schneider_Klenner_Karl_2020, title={Every Node for Itself:
    Fully Distributed Service Coordination}, booktitle={IEEE International Conference
    on Network and Service Management (CNSM)}, publisher={IEEE}, author={Schneider,
    Stefan Balthasar and Klenner, Lars Dietrich and Karl, Holger}, year={2020} }'
  chicago: 'Schneider, Stefan Balthasar, Lars Dietrich Klenner, and Holger Karl. “Every
    Node for Itself: Fully Distributed Service Coordination.” In <i>IEEE International
    Conference on Network and Service Management (CNSM)</i>. IEEE, 2020.'
  ieee: 'S. B. Schneider, L. D. Klenner, and H. Karl, “Every Node for Itself: Fully
    Distributed Service Coordination,” in <i>IEEE International Conference on Network
    and Service Management (CNSM)</i>, 2020.'
  mla: 'Schneider, Stefan Balthasar, et al. “Every Node for Itself: Fully Distributed
    Service Coordination.” <i>IEEE International Conference on Network and Service
    Management (CNSM)</i>, IEEE, 2020.'
  short: 'S.B. Schneider, L.D. Klenner, H. Karl, in: IEEE International Conference
    on Network and Service Management (CNSM), IEEE, 2020.'
date_created: 2020-09-22T06:23:40Z
date_updated: 2022-01-06T06:54:08Z
ddc:
- '006'
department:
- _id: '75'
file:
- access_level: open_access
  content_type: application/pdf
  creator: stschn
  date_created: 2020-09-22T06:25:57Z
  date_updated: 2020-09-22T06:36:25Z
  file_id: '19608'
  file_name: ris_with_copyright.pdf
  file_size: 500948
  relation: main_file
file_date_updated: 2020-09-22T06:36:25Z
has_accepted_license: '1'
keyword:
- distributed management
- service coordination
- network coordination
- nfv
- softwarization
- orchestration
language:
- iso: eng
oa: '1'
project:
- _id: '1'
  name: SFB 901
- _id: '4'
  name: SFB 901 - Project Area C
- _id: '16'
  name: SFB 901 - Subproject C4
publication: IEEE International Conference on Network and Service Management (CNSM)
publisher: IEEE
status: public
title: 'Every Node for Itself: Fully Distributed Service Coordination'
type: conference
user_id: '35343'
year: '2020'
...
---
_id: '19609'
abstract:
- lang: eng
  text: "Modern services comprise interconnected components,\r\ne.g., microservices
    in a service mesh, that can scale and\r\nrun on multiple nodes across the network
    on demand. To process\r\nincoming traffic, service components have to be instantiated
    and\r\ntraffic assigned to these instances, taking capacities and changing\r\ndemands
    into account. This challenge is usually solved with\r\ncustom approaches designed
    by experts. While this typically\r\nworks well for the considered scenario, the
    models often rely\r\non unrealistic assumptions or on knowledge that is not available\r\nin
    practice (e.g., a priori knowledge).\r\n\r\nWe propose a novel deep reinforcement
    learning approach that\r\nlearns how to best coordinate services and is geared
    towards\r\nrealistic assumptions. It interacts with the network and relies on\r\navailable,
    possibly delayed monitoring information. Rather than\r\ndefining a complex model
    or an algorithm how to achieve an\r\nobjective, our model-free approach adapts
    to various objectives\r\nand traffic patterns. An agent is trained offline without
    expert\r\nknowledge and then applied online with minimal overhead. Compared\r\nto
    a state-of-the-art heuristic, it significantly improves flow\r\nthroughput and
    overall network utility on real-world network\r\ntopologies and traffic traces.
    It also learns to optimize different\r\nobjectives, generalizes to scenarios with
    unseen, stochastic traffic\r\npatterns, and scales to large real-world networks."
author:
- first_name: Stefan Balthasar
  full_name: Schneider, Stefan Balthasar
  id: '35343'
  last_name: Schneider
  orcid: 0000-0001-8210-4011
- first_name: Adnan
  full_name: Manzoor, Adnan
  last_name: Manzoor
- first_name: Haydar
  full_name: Qarawlus, Haydar
  last_name: Qarawlus
- first_name: Rafael
  full_name: Schellenberg, Rafael
  last_name: Schellenberg
- first_name: Holger
  full_name: Karl, Holger
  id: '126'
  last_name: Karl
- first_name: Ramin
  full_name: Khalili, Ramin
  last_name: Khalili
- first_name: Artur
  full_name: Hecker, Artur
  last_name: Hecker
citation:
  ama: 'Schneider SB, Manzoor A, Qarawlus H, et al. Self-Driving Network and Service
    Coordination Using Deep Reinforcement Learning. In: <i>IEEE International Conference
    on Network and Service Management (CNSM)</i>. IEEE; 2020.'
  apa: Schneider, S. B., Manzoor, A., Qarawlus, H., Schellenberg, R., Karl, H., Khalili,
    R., &#38; Hecker, A. (2020). Self-Driving Network and Service Coordination Using
    Deep Reinforcement Learning. In <i>IEEE International Conference on Network and
    Service Management (CNSM)</i>. IEEE.
  bibtex: '@inproceedings{Schneider_Manzoor_Qarawlus_Schellenberg_Karl_Khalili_Hecker_2020,
    title={Self-Driving Network and Service Coordination Using Deep Reinforcement
    Learning}, booktitle={IEEE International Conference on Network and Service Management
    (CNSM)}, publisher={IEEE}, author={Schneider, Stefan Balthasar and Manzoor, Adnan
    and Qarawlus, Haydar and Schellenberg, Rafael and Karl, Holger and Khalili, Ramin
    and Hecker, Artur}, year={2020} }'
  chicago: Schneider, Stefan Balthasar, Adnan Manzoor, Haydar Qarawlus, Rafael Schellenberg,
    Holger Karl, Ramin Khalili, and Artur Hecker. “Self-Driving Network and Service
    Coordination Using Deep Reinforcement Learning.” In <i>IEEE International Conference
    on Network and Service Management (CNSM)</i>. IEEE, 2020.
  ieee: S. B. Schneider <i>et al.</i>, “Self-Driving Network and Service Coordination
    Using Deep Reinforcement Learning,” in <i>IEEE International Conference on Network
    and Service Management (CNSM)</i>, 2020.
  mla: Schneider, Stefan Balthasar, et al. “Self-Driving Network and Service Coordination
    Using Deep Reinforcement Learning.” <i>IEEE International Conference on Network
    and Service Management (CNSM)</i>, IEEE, 2020.
  short: 'S.B. Schneider, A. Manzoor, H. Qarawlus, R. Schellenberg, H. Karl, R. Khalili,
    A. Hecker, in: IEEE International Conference on Network and Service Management
    (CNSM), IEEE, 2020.'
date_created: 2020-09-22T06:28:22Z
date_updated: 2022-01-06T06:54:08Z
ddc:
- '006'
department:
- _id: '75'
file:
- access_level: open_access
  content_type: application/pdf
  creator: stschn
  date_created: 2020-09-22T06:29:16Z
  date_updated: 2020-09-22T06:36:00Z
  file_id: '19610'
  file_name: ris_with_copyright.pdf
  file_size: 642999
  relation: main_file
file_date_updated: 2020-09-22T06:36:00Z
has_accepted_license: '1'
keyword:
- self-driving networks
- self-learning
- network coordination
- service coordination
- reinforcement learning
- deep learning
- nfv
language:
- iso: eng
oa: '1'
project:
- _id: '1'
  name: SFB 901
- _id: '4'
  name: SFB 901 - Project Area C
- _id: '16'
  name: SFB 901 - Subproject C4
publication: IEEE International Conference on Network and Service Management (CNSM)
publisher: IEEE
status: public
title: Self-Driving Network and Service Coordination Using Deep Reinforcement Learning
type: conference
user_id: '35343'
year: '2020'
...