본문

서브메뉴

Assisting Cloud System Development With Automated Insight Generation
Assisting Cloud System Development With Automated Insight Generation
Assisting Cloud System Development With Automated Insight Generation

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20260202103648
ISBN  
9798314875407
DDC  
004
저자명  
Qiu, Yiming.
서명/저자  
Assisting Cloud System Development With Automated Insight Generation
발행사항  
[Sl] : University of Michigan, 2025
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2025
형태사항  
141 p
주기사항  
Source: Dissertations Abstracts International, Volume: 86-11, Section: B.
주기사항  
Advisor: Chen, Ang.
학위논문주기  
Thesis (Ph.D.)--University of Michigan, 2025.
초록/해제  
요약As cloud computing revolutionizes the IT industry, the complexity of underlying cloud systems has been increasing rapidly. The advent of new hardware accelerators and software platforms has made it challenging for cloud users to master the growing development toolkits. Compounding the issue, the programming frameworks and internals of these new systems are highly heterogeneous, with different performance characteristics, resource constraints, management principles, and reliability considerations. Consequently, it is becoming crucial to minimize human effort when managing these new ecosystems. In this dissertation, we advocate for assisting cloud developers and operators by automatically generating system insights. These insights bridge the gap between user intentions and system requirements, providing clarity on the outcomes of user actions on a system without the need for tedious trial-and-error processes.This dissertation demonstrates how we generate various types of insights for different cloud systems. Firstly, the dissertation explores performance optimization insights, which are critically needed as users attempt to offload legacy code from on-premise servers to emerging accelerators like SmartNICs. These new hardware components feature entirely different programming abstractions, compilers, instruction sets, and architectures. Although a straightforward offloading strategy might functionally work, it could lead to significant performance degradation, undermining the benefits of using accelerators. To address this issue, we create a toolset called Clara, which can automatically predict offloading performance and suggest tuning strategies before extensive deployment efforts. This allows users to make informed decisions on whether and how to offload their legacy code. Secondly, the dissertation investigates safety compliance insights for the cloud networking stack, focusing on ensuring the correctness of system updates for the latest generation of runtime-programmable platforms. We observe that even if both the current and intended functionalities are correct and efficient, the intermediate transition state can still introduce consistency and capacity issues into the core network. To tackle this challenge, we employ formal reasoning techniques to achieve update clarity. We develop FlexPlan, an interactive platform that synthesizes runtime transition plans meeting dynamic user demands, greatly minimizing the need for manual intervention.Lastly, the dissertation unearths infrastructure management insights for emerging cloud orchestration platforms. Clouds are constructed by providers like Microsoft but are intended for third-party use. This user/owner division limits cloud users' visibility and control over cloud service behavior. The adoption of Infrastructure-as-Code (IaC) style cloud orchestration platforms further complicates this semantic gap by adding another intermediate layer of abstraction. To address this complexity, we propose Zodiac, a pipeline that automatically uncovers cloud provider requirements, and clarifies their interaction with orchestration platforms. The outcome is a set of orchestration rules that cloud users must follow to ensure proper cloud management practices.Throughout these projects, we leverage and extend techniques from a wide variety of disciplines, such as formal reasoning, software testing, machine learning, and their intersections. The results demonstrate the feasibility of generating useful insights across cloud data, control, and management planes, while unveiling an even larger insight generation and integration design space yet to be explored. 
일반주제명  
Computer science
일반주제명  
Computer engineering
일반주제명  
Information technology
키워드  
Cloud management
키워드  
Programmable network
키워드  
Program analysis
키워드  
Program synthesis
키워드  
Machine learning
키워드  
Configuration mining
기타저자  
University of Michigan Computer Science & Engineering
기본자료저록  
Dissertations Abstracts International. 86-11B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260126s2025        us                              c    eng  d
■001000017358128
■00520260202103648
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798314875407
■035    ▼a(MiAaPQ)AAI32092654
■035    ▼a(MiAaPQ)umichrackham006183
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a004
■1001  ▼aQiu,  Yiming.
■24510▼aAssisting  Cloud  System  Development  With  Automated  Insight  Generation
■260    ▼a[Sl]▼bUniversity  of  Michigan▼c2025
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2025
■300    ▼a141  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  86-11,  Section:  B.
■500    ▼aAdvisor:  Chen,  Ang.
■5021  ▼aThesis  (Ph.D.)--University  of  Michigan,  2025.
■520    ▼aAs  cloud  computing  revolutionizes  the  IT  industry,  the  complexity  of  underlying  cloud  systems  has  been  increasing  rapidly.  The  advent  of  new  hardware  accelerators  and  software  platforms  has  made  it  challenging  for  cloud  users  to  master  the  growing  development  toolkits.  Compounding  the  issue,  the  programming  frameworks  and  internals  of  these  new  systems  are  highly  heterogeneous,  with  different  performance  characteristics,  resource  constraints,  management  principles,  and  reliability  considerations.  Consequently,  it  is  becoming  crucial  to  minimize  human  effort  when  managing  these  new  ecosystems.  In  this  dissertation,  we  advocate  for  assisting  cloud  developers  and  operators  by  automatically  generating  system  insights.  These  insights  bridge  the  gap  between  user  intentions  and  system  requirements,  providing  clarity  on  the  outcomes  of  user  actions  on  a  system  without  the  need  for  tedious  trial-and-error  processes.This  dissertation  demonstrates  how  we  generate  various  types  of  insights  for  different  cloud  systems.  Firstly,  the  dissertation  explores  performance  optimization  insights,  which  are  critically  needed  as  users  attempt  to  offload  legacy  code  from  on-premise  servers  to  emerging  accelerators  like  SmartNICs.  These  new  hardware  components  feature  entirely  different  programming  abstractions,  compilers,  instruction  sets,  and  architectures.  Although  a  straightforward  offloading  strategy  might  functionally  work,  it  could  lead  to  significant  performance  degradation,  undermining  the  benefits  of  using  accelerators.  To  address  this  issue,  we  create  a  toolset  called  Clara,  which  can  automatically  predict  offloading  performance  and  suggest  tuning  strategies  before  extensive  deployment  efforts.  This  allows  users  to  make  informed  decisions  on  whether  and  how  to  offload  their  legacy  code. Secondly,  the  dissertation  investigates  safety  compliance  insights  for  the  cloud  networking  stack,  focusing  on  ensuring  the  correctness  of  system  updates  for  the  latest  generation  of  runtime-programmable  platforms.  We  observe  that  even  if  both  the  current  and  intended  functionalities  are  correct  and  efficient,  the  intermediate  transition  state  can  still  introduce  consistency  and  capacity  issues  into  the  core  network.  To  tackle  this  challenge,  we  employ  formal  reasoning  techniques  to  achieve  update  clarity.  We  develop  FlexPlan,  an  interactive  platform  that  synthesizes  runtime  transition  plans  meeting  dynamic  user  demands,  greatly  minimizing  the  need  for  manual  intervention.Lastly,  the  dissertation  unearths  infrastructure  management  insights  for  emerging  cloud  orchestration  platforms.  Clouds  are  constructed  by  providers  like  Microsoft  but  are  intended  for  third-party  use.  This  user/owner  division  limits  cloud  users'  visibility  and  control  over  cloud  service  behavior.  The  adoption  of  Infrastructure-as-Code  (IaC)  style  cloud  orchestration  platforms  further  complicates  this  semantic  gap  by  adding  another  intermediate  layer  of  abstraction.  To  address  this  complexity,  we  propose  Zodiac,  a  pipeline  that  automatically  uncovers  cloud  provider  requirements,  and  clarifies  their  interaction  with  orchestration  platforms.  The  outcome  is  a  set  of  orchestration  rules  that  cloud  users  must  follow  to  ensure  proper  cloud  management  practices.Throughout  these  projects,  we  leverage  and  extend  techniques  from  a  wide  variety  of  disciplines,  such  as  formal  reasoning,  software  testing,  machine  learning,  and  their  intersections.  The  results  demonstrate  the  feasibility  of  generating  useful  insights  across  cloud  data,  control,  and  management  planes,  while  unveiling  an  even  larger  insight  generation  and  integration  design  space  yet  to  be  explored. 
■590    ▼aSchool  code:  0127.
■650  4▼aComputer  science
■650  4▼aComputer  engineering
■650  4▼aInformation  technology
■653    ▼aCloud  management
■653    ▼aProgrammable  network
■653    ▼aProgram  analysis
■653    ▼aProgram  synthesis
■653    ▼aMachine  learning
■653    ▼aConfiguration  mining
■690    ▼a0984
■690    ▼a0489
■690    ▼a0464
■690    ▼a0800
■71020▼aUniversity  of  Michigan▼bComputer  Science  &  Engineering.
■7730  ▼tDissertations  Abstracts  International▼g86-11B.
■790    ▼a0127
■791    ▼aPh.D.
■792    ▼a2025
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17358128▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF16008 전자도서 대출가능 마이폴더 부재도서신고 비도서대출신청 야간 도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.