Service management in NoSQL data stores via replica-group reconfigurations (Master thesis)
Internet-scale services increasingly rely on NoSQL data store technologies for scalable, highly available data persistence. Workload variations or imbalances in data-store resource utilization typically require adaptation actions that involve movement of data (e.g., migrating data between nodes) and thus involve high overhead. In this thesis we propose the use of low-cost adaptation actions based on targeted reconfigurations of replica groups as an effective way to offer rapid response to performance degradation (such as when a node experiences a temporary resource shortage). Such reconfigurations can be exercised and controlled by a management system that uses measurement-based analysis to ensure they are only applied when their benefit outweighs their cost. Such a management system may also leverage external cross-layer notifications from an underlying container management system (CMS) about resourcespecific conditions that are expected to impact the data store. We evaluate a prototype implementation of the management system in the context of the RethinkDB data store on Google Container Engine with the Kubernetes CMS, and on a dedicated cluster, using the Yahoo Cloud Serving Benchmark. Our results demonstrate that low-cost data-store adaptation actions can improve overall system manageability, availability, and performance, and that events exposed by the CMS can be leveraged to drive proactive adaptation actions, improving overall quality of service.
|Institution and School/Department of submitter:||Πανεπιστήμιο Ιωαννίνων. Σχολή Θετικών Επιστημών. Τμήμα Μηχανικών Η/Υ & Πληροφορικής|
|Subject classification:||Service management|
|Keywords:||Αποθετήρια δεδομένων,Αυτόνομη διαχείριση,Αναδιοργανώσεις ομάδων αντιγράφων,Διαστρωμάτωση,Autonomic management,NoSQL datastores,Cross-layer,Reconfigurations|
|Appears in Collections:||Διατριβές Μεταπτυχιακής Έρευνας (Masters)|
Files in This Item:
|Μ.Ε. ΜΠΕΚΑΣ ΕΥΔΟΞΟΣ 2017.pdf||2.87 MB||Adobe PDF||View/Open|
Please use this identifier to cite or link to this item:This item is a favorite for 0 people.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.