Uporaba avtomatizacije za doseganje visoko razpoložljivega podatkovnega centra

Lulik, Matic (author), Sedlar, Urban (mentor)

V magistrskem delu je predstavljena postavitev razvojnega okolja za testiranje avtomatizacije za doseganje visoko razpoložljivega podatkovnega centra. Opisan je razvoj avtomatizacije, ki sem ga za namen izdelave magistrske naloge delno razvil sam. Magistrska naloga je moje delo, rešitve, uporabljene v magistrskem delu, pa so bile ustvarjene ob sodelovanju z drugimi sodelavci v podjetju, kjer sem zaposlen, s katerimi smo razvijali rešitev za eno izmed naših strank. Ob tem lahko poudarim, da se rešitev, predstavljena v nalogi, uspešno uporablja v produkcijskem okolju stranke. V prvem delu naloge so omenjene in razložene dobre prakse podatkovnih centrov za doseganje visoke razpoložljivosti. Prav tako je opisana razlika med preprečevanjem katastrof v podatkovnem centru (ang. {Disaster Prevention) in obnovitvijo podatkov oz. infrastrukture po katastrofi (ang. Disaster Recovery). Omenjeni so nekateri izmed načinov za doseganje zmožnosti obnovitve sistemov ob neželenih dogodkih oz. razlogi za doseganje visoke razpoložljivosti podatkovnih centrov. V vsakem primeru je bolje, da do izpada ne pride, zato je v nadaljevanju predstavljena razvita rešitev, namenjena preprečitvi izpada oz. katastrofe (ang. Disaster Prevention). Opisani so gradniki razvojnega okolja in avtomatizacije, ki je bila testirana in razvita na razvojnem okolju. Gradniki razvojnega okolja in gradniki, na katerih temelji implementirana rešitev, so v glavnem produkti proizvajalca VMware in Cisco, za avtomatizacijo pa je uporabljeno orodje Ansible. Vsi gradniki so podrobneje opisani v nadaljevanju. V zadnjem delu magistrske naloge so bili na razvojnem okolju izvedeni testi, v katerih sem meril čas nedosegljivosti omrežnih prehodov, navideznih strojev in spletnih aplikacij pri migraciji gradnikov v primeru katastrofe iz enega na drugi podatkovni center. Po pričakovanjih je migracija navideznih strojev z uporabo avtomatizacije hitrejša, zanesljivejša in natančnejša v primerjavi z ročnim načinom migracije. S testi in meritvami je bilo ugotovljeno, da so z uporabo avtomatizacije omrežja nedosegljiva v povprečju 2 min in 19 s manj, navidezni stroji 5 min 41 s manj, spletne aplikacije pa 4 min 15 s manj kot pri ročnem načinu migracije. Prav tako administrator pri ročnem načinu migracije porabi 5 min in 16 s več kot skripta pri avtomatizaciji.

• Language: Slovenian
• Keywords: Preprečitev katastrof, obnovitev po katastrofi, visoka razpoložljivost, avtomatizacija, Cisco, VMware, NSX
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FE - Faculty of Electrical Engineering
• Year: 2024
• PID: 20.500.12556/RUL-162991
• COBISS.SI-ID: 215727107
• Publication date in RUL: 30.09.2024

Secondary language

• Language: English
• Title: Using Automation to Achieve a Highly Available Data Center

Abstract

The master's thesis presents the setup of a development environment for testing automation to achieve a highly available data center. It describes the development of automation, which was partially developed by me for the purpose of completing the thesis. While the thesis is my work, the solutions presented in it were created in collaboration with colleagues from the company where I am employed, as we developed the solution for one of our clients. This highlights that the solution presented in the thesis is successfully used in a production environment and has already been successfully applied in certain situations. The first part of the document mentions and explains best practices in data centers to achieve high availability. It also describes the difference between disaster prevention in a data center and disaster recovery. Some of the methods for achieving system recovery capabilities in the event of unwanted incidents, as well as the reasons for achieving high availability in data centers, are discussed. In any case, it is preferable to avoid outages altogether. The document presents a developed solution aimed at preventing outages or disasters (disaster prevention). The components of the development environment and automation, which were tested and developed in the development environment, are described. The components of the development environment and those on which the implemented solution is based are mainly products from VMware and Cisco, while Ansible is used for automation. All components are described in detail. In the final part of the thesis, tests were conducted in the development environment, where I measured the downtime of network gateways, virtual machines, and web applications during migration in the event of a disaster from one data center to another. As expected, the migration of virtual machines using automation was faster, more reliable, and more precise compared to manual migration. The tests and measurements showed that, by using automation, network gateways were on average unreachable for 2 minutes and 19 seconds less, virtual machines for 5 minutes and 41 seconds less, and web applications for 4 minutes and 15 seconds less than with manual migration. Furthermore, an administrator performing manual migration required on average 5 minutes and 16 seconds more than the automated script.

• Keywords: Disaster Prevention, Disaster Recovery, High Availability, Automation, Cisco, VMware, NSX