The steady growth of cloud computing platforms has caused a major interest towards the development of systems capable of collecting data in a distributed way and resulting tolerant to failures. In this graduation thesis the data storage process is analyzed by means of a distributed storage system of latest generation denominated Ceph. Its application is chiefly oriented to cloud computing platforms as OpenStack. This degree thesis can be divided in three different parts related to the environment of the storage platform, the explanation of its structure and finally its benchmarks. First of all, some documents concerning the structure and components of OpenStack were analyzed in order to be used in conjunction with Ceph; hereafter a software version was installed in way of testing the combined operations. Stated the innovation and the promises of the previously mentioned system, it was decided to deepen the study; actually the only cloud storage platform was installed on different virtual machines with the purpose of resulting similar to a real system to test the operation of individual components through focused benchmarks. During the platform employment an erasure code method was discovered: it is still developing with Fujitsu Laboratories and it is not promoted yet; therefore an algorithm was implemented to be able to verify the encoding and decoding speed compared with more traditional methods of forward error correction. Thanks to its innovative structure, the cloud storage system that will be studied and illustrated in this graduation thesis is capable of not having bottlenecks and single points of failure; it is also able to be install on machines with common-hardware through the workload balance and the saved data making it perhaps the best choice that can be carried out in this field.
La costante espansione delle piattaforme di cloud computing ha portato un maggior interesse verso lo sviluppo di sistemi in grado di raccogliere dati in modo distribuito e risultare tolleranti a fallimenti. In questa tesi viene analizzato il processo di immagazzinamento di dati attraverso un sistema di storage distribuito di ultima generazione denominato Ceph. Il suo utilizzo è prevalentemente rivolto alle piattaforme di cloud computing come OpenStack. La tesi si può suddividere in tre parti differenti che riguardano l’ambiente di utilizzo della piattaforma di storage, la spiegazione della sua struttura e per ultimo i benchmark della stessa. In primo luogo si sono analizzati i documenti riguardanti la struttura e i componenti di OpenStack per essere utilizzata in concomitanza di Ceph; in seguito si è installata una versione dei software in modo da testarne il funzionamento congiunto. Data l’innovazione e le promesse dell’ultimo sistema citato, si è scelto di approfondirne lo studio; si è infatti installata la sola piattaforma di cloud storage su più macchine virtuali in maniera tale da risultare simile ad un sistema reale per sottoporre a verifica il funzionamento dei singoli componenti tramite dei benchmark mirati. Durante l’utilizzo della piattaforma si è poi scoperto un metodo di erasure code ancora in via di sviluppo con i laboratori Fujitsu e non pubblicizzato; di conseguenza si è sviluppato un algoritmo in grado di controllarne la velocità di encoding e di decoding rispetto ai metodi di correzione degli errori più “tradizionali”. Il sistema di cloud storage che verrà studiato e descritto in questa tesi, grazie alla sua innovativa struttura è in grado di non avere colli di bottiglia e unici punti di fallimento, riuscendo inoltre a essere installato su macchine con hardware comune tramite il bilanciamento del carico di lavoro e dei dati salvati, rendendolo di fatto forse la miglior scelta che si possa effettuare in questo campo.
Cloud storage : analisi e documentazione attraverso benchmark specifici di Ceph
GASPARETTI, JACOPO;LOREGIOLA, ALESSANDRO
2014/2015
Abstract
The steady growth of cloud computing platforms has caused a major interest towards the development of systems capable of collecting data in a distributed way and resulting tolerant to failures. In this graduation thesis the data storage process is analyzed by means of a distributed storage system of latest generation denominated Ceph. Its application is chiefly oriented to cloud computing platforms as OpenStack. This degree thesis can be divided in three different parts related to the environment of the storage platform, the explanation of its structure and finally its benchmarks. First of all, some documents concerning the structure and components of OpenStack were analyzed in order to be used in conjunction with Ceph; hereafter a software version was installed in way of testing the combined operations. Stated the innovation and the promises of the previously mentioned system, it was decided to deepen the study; actually the only cloud storage platform was installed on different virtual machines with the purpose of resulting similar to a real system to test the operation of individual components through focused benchmarks. During the platform employment an erasure code method was discovered: it is still developing with Fujitsu Laboratories and it is not promoted yet; therefore an algorithm was implemented to be able to verify the encoding and decoding speed compared with more traditional methods of forward error correction. Thanks to its innovative structure, the cloud storage system that will be studied and illustrated in this graduation thesis is capable of not having bottlenecks and single points of failure; it is also able to be install on machines with common-hardware through the workload balance and the saved data making it perhaps the best choice that can be carried out in this field.| File | Dimensione | Formato | |
|---|---|---|---|
|
2015_10_Gasparetti_Loregiola.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
6.95 MB
Formato
Adobe PDF
|
6.95 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/112566