Fault-tolerant virtualization
| Název práce v češtině: | Virtualizace odolná vůči chybám |
|---|---|
| Název v anglickém jazyce: | Fault-tolerant virtualization |
| Klíčová slova: | virtualizace, odolnost vůči chybám, QEMU |
| Klíčová slova anglicky: | virtualization, fault-tolerant, QEMU |
| Akademický rok vypsání: | 2013/2014 |
| Typ práce: | diplomová práce |
| Jazyk práce: | angličtina |
| Ústav: | Katedra softwarového inženýrství (32-KSI) |
| Vedoucí / školitel: | RNDr. Jakub Yaghob, Ph.D. |
| Řešitel: | skrytý - zadáno a potvrzeno stud. odd. |
| Datum přihlášení: | 11.10.2013 |
| Datum zadání: | 08.11.2013 |
| Datum potvrzení stud. oddělením: | 18.11.2013 |
| Datum a čas obhajoby: | 08.09.2014 09:30 |
| Datum odevzdání elektronické podoby: | 29.07.2014 |
| Datum odevzdání tištěné podoby: | 30.07.2014 |
| Datum proběhlé obhajoby: | 08.09.2014 |
| Oponenti: | Mgr. Martin Děcký, Ph.D. |
| Zásady pro vypracování |
| Historically, service downtime, caused by hardware failure, could either be avoided by using an expensive fault-tolerant hardware, or by using an in-service clustering. The first approach is not applicable in most cases due to very high cost of such hardware, whereas the second approach often requires extensive changes to the actual service application. There is, however, one more approach. One can use hardware virtualization to provide fault tolerance transparently to any service. The aim of this thesis is to create such a system, and evaluate performace, comparing it to non-fault-tolerant environment.
The student should research possible techniques for using virtualization technologies to provide a fault tolerant system. To prove their understanding, they should implement at least one such technique in an open-source hypervisor (for example Qemu-KVM). The student should also create a set of workloads, consisting real-world use cases of virtual machines, and test the performance hit of using a fault tolerant virtualization, as opposed to normal virtualization. This benchmark should be done on both the modified hypervisor, and at least one commercially available product (such as VMWare ESXi). Lastly, the student should investigate the techniques used in benchmarked products, and create a second set of workloads, designed specifically to have the highest performance hit when combined with said fault tolerancy techniques. These workloads should be benchmarked as well, to show what is the performance hit in worst case scenario. As a possible extended exercise, the student would test the impact of different server interconnect technologies (some of the following: FastEthernet, GigabitEthernet, 10GEthernet, InfiniBand) on the benchmarking result, where possible. |
| Seznam odborné literatury |
| Robert Hanmer: Patterns for Fault Tolerant Software
Kenneth Hess, Amy Newman: Practical Virtualization Solutions: Virtualization from the Trenches Duncan Epping, Frank Denneman - VMware vSphere 5.1 Clustering Deepdive Edward Haletky: VMware ESX and ESXi in the Enterprise Robert Warnke, Thomas Ritzau: qemu-kvm & libvirt |