Abstract
dc:descriptionIn modern cloud environments, virtualization enables multiple tenants to flexibly share the datacenter manycore machines with high processing and memory capacity. Operating-system containers are a lightweight form of virtualization that is commonly used to run the data-intensive applications of different tenants in cloud infrastructures, because they offer flexible virtualization with low overhead. At a high level, two main reasons make this sharing challenging. First, the system kernel of a cloud machine serving software containers can become a performance bottleneck and an attack surface for the colocated tenants. Second, the secure management of the enormous user populations that belong to different tenants in a cloud environment necessitates the design of scalable and tenant-aware access control policies at the infrastructure-level. The focus of this dissertation is to enable multiple tenants to efficiently and securely share the computing, storage, and network infrastructure of the datacenter. To this end, we take the radical approach of moving the data-intensive I/O services at user level from the shared kernel, in order to serve the containers of competing tenants over the same cloud machines. Our contributions consist of innovative methods to handle POSIX-like system calls at user level through a library, the producer-consumer transfer of data and requests over shared memory with efficient memory copy and relaxed lock-free queues, the construction of stacked user-level I/O services, and a multitenant access control mechanism built natively into a distributed filesystem. This dissertation shows that it is possible to allow the data-intensive applications of different tenants to share the same datacenter machines with reduced contention for shared resources. The systems that we propose enable the cloud operation with lower cost, more predictable I/O latency and throughput, increased resilience to attacks from colocated tenants, increased performance for the end users under contention conditions, improved energy efficiency overall as a result of the better resource utilization achieved.
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kappes, Georgios
- Καππές, Γεώργιος
Subjects
dc:subject × 36- Υπολογιστικό νέφος
- Εικονικοποίηση
- Περιέκτες λογισμικού
- Αποθήκευση δεδομένων
- Κατανεμημένα συστήματα αρχείων
- Ταυτόχρονες Δομές Δεδομένων
- Ουρές
- Υπηρεσίες επιπέδου χρήστη
- Ασφάλεια υπολογιστικού νέφους
- Έλεγχος πρόσβασης
- Πολυμισθωτικότητα
- Αντιγραφή μνήμης
- Cloud computing
- Virtualization
- Software containers
- Data storage
- Distributed file systems
- Concurrent Data Structures
- Queues
- User-level services
- Cloud Security
- Access control
- Multitenancy
- Memory copy
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Πληροφοριακά συστήματα
- Επιστήμη ηλεκτρονικών υπολογιστών
- Λογισμικό (software)
- Επιστήμη ηλεκτρονικών υπολογιστών, θεωρία και μέθοδοι
- Computer and Information Sciences
- Natural Sciences
- Information Systems
- Computer Science
- Software
- Computer science, theory and methods
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/49593