Evaluating the cost and performance of cloud on-demand services against in-house smb-virtualisation resources
Toivanen, Teemu (2016-09-07)
Evaluating the cost and performance of cloud on-demand services against in-house smb-virtualisation resources
Toivanen, Teemu
(07.09.2016)
Tätä artikkelia/julkaisua ei ole tallennettu UTUPubiin. Julkaisun tiedoissa voi kuitenkin olla linkki toisaalle tallennettuun artikkeliin / julkaisuun.
Turun yliopisto
Kuvaus
Siirretty Doriasta
Tiivistelmä
Cloud computing is an emerging commercial IT infrastructure paradigm. Contrary to the traditional paradigm where hardware needs to be acquired upfront, in cloud computing the services are managed and delivered over the internet. Cloud computing enables organisations to outsource their computing resource requirements to gain greater reliability and cost savings. Cloud computing services are provided on demand and are classied into three fundamental models: Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS), based on the level of abstraction. This thesis focuses on the IaaS model cloud computing service and compares the performance and costs of three widely-used public cloud computing service providers (Amazon, Google and Microsoft) to a small on-premises private cloud platform.
The study focuses on benchmarking of the IaaS cloud from a user's perspective. It is important for organisations to make the right decision before shifting their business to the cloud. In particular, it should be ensured that the applications which have been moved to the cloud will be resilient and meet the predened performance Service Level Agreements(SLAs). The present study focuses on the comparison of computational and disk resources and costs of the dierent service providers. To avoid cloud provider bias and benchmark program version dierences all test instances were created from an identical Virtual Disk Operating System (VDOS) image.
Instance machine types for the tests are low-end and mid-range machine resources with dierent disk back-ends where applicable. The study uses well known benchmark tests. The costs of using the commercial services limited the number of repetitions in the present work.
All tests for the similar machine types were started at the same time. The tests were managed and results collected from a central management system outside of the cloud computing systems.
The present study proposes a way to compare the services of cloud providers. The results show that there can be variance between instances in the same providers cloud. The total cost for using cloud instances is sum of instance type, uptime, network trac, disk usage, and other cost factors that are cloud provider specic. This makes the comparison of cloud provider value for money dicult and application dependent. On the other hand, using an on-premises self-hosted private cloud can be tailored to the needs of the application, but this requires upfront expenditure for hardware.
The study focuses on benchmarking of the IaaS cloud from a user's perspective. It is important for organisations to make the right decision before shifting their business to the cloud. In particular, it should be ensured that the applications which have been moved to the cloud will be resilient and meet the predened performance Service Level Agreements(SLAs). The present study focuses on the comparison of computational and disk resources and costs of the dierent service providers. To avoid cloud provider bias and benchmark program version dierences all test instances were created from an identical Virtual Disk Operating System (VDOS) image.
Instance machine types for the tests are low-end and mid-range machine resources with dierent disk back-ends where applicable. The study uses well known benchmark tests. The costs of using the commercial services limited the number of repetitions in the present work.
All tests for the similar machine types were started at the same time. The tests were managed and results collected from a central management system outside of the cloud computing systems.
The present study proposes a way to compare the services of cloud providers. The results show that there can be variance between instances in the same providers cloud. The total cost for using cloud instances is sum of instance type, uptime, network trac, disk usage, and other cost factors that are cloud provider specic. This makes the comparison of cloud provider value for money dicult and application dependent. On the other hand, using an on-premises self-hosted private cloud can be tailored to the needs of the application, but this requires upfront expenditure for hardware.