In this thesis, there are discussed use-cases of NoSQL database MongoDB implemented as a datastore for user data, which is stored by Datastore stubs in Google App Engine SDK. Existing stubs are not very well optimized for higher load; they significantly slow down application development and testing if there is a need to store larger data sets in these storages. The analysis is focused on features of MongoDB, Google App Engine NoSQL Datastore and interfaces for data manipulation in SDK - Datastore Service Stub API. As a result, there was designed and implemented new datastore stub, which is supposed to solve problems of existing stubs. New stub uses MongoDB as a database layer for storing testing data and it is fully integrated into Google App Engine SDK.

Consistency preservation of replicated data is a critical aspect for distributed databaseswhich are strongly consistent. Further, in fail-recovery model each process also needs todeal with the management of stable storage and amnesia [1]. CATS is a key/value datastore which combines the Distributed Hash Table (DHT) like scalability and selforganization and also provides atomic consistency of the replicated items. However beingan in memory data store with consistency and partition tolerance (CP), it suffers frompermanent unavailability in the event of majority failure. The goals of this thesis were twofold (i) to implement disk persistent storage in CATS,which would allow the records and state of the nodes to be persisted on disk and (ii) todesign nodes failure recovery-algorithm for CATS which enable the system to run with theassumption of a Fail Recovery model without violating consistency. For disk persistent storage two existing key/value databases LevelDB [2] and BerkleyDB[3] are used. LevelDB is an implementation of log structured merged trees [4] where asBerkleyDB is an implementation of log structured B+ trees [5]. Both have been used as anunderlying local storage for nodes and throughput and latency of the system with each isdiscussed. A technique to improve the performance by allowing concurrent operations onthe nodes is also discussed. The nodes failure-recovery algorithm is designed with a goalto allow the nodes to crash and then recover without violating consistency and also toreinstate availability once the majority of nodes recover. The recovery algorithm is based onpersisting the state variables of Paxos [6] acceptor and proposer and consistent groupmemberships. For fault-tolerance and recovery, processes also need to copy records from the replicationgroup. This becomes problematic when the number of records and the amount of data ishuge. For this problem a technique for transferring key/value records in bulk is alsodescribed, and its effect on the latency and throughput of the system is discussed.

I takt med att datakvantiteter växer och kraven på skalbarhet och tillgänglighet inom molntjänster växer, framhävs behovet av undersökningar kring dess prestanda och kostnadseffektivitet.  Dessa analyser är avgörande för att optimera tjänster och bistå företag med värdefulla rekommendationer för att fatta välgrundade beslut om datalagring i molnet. Detta examensarbete undersöker kostnad samt prestanda hos relationella och icke-relationella datalagringslösningar implementerade på Microsoft Azure och Google Cloud Platform. Verktyget Hyperfine används för att mäta latens och tjänsternas kostnadseffektivitet beräknas baserat på detta resultat samt dess beräknade månadskostnader. Studiens resultat indikerar att för de utvärderade relationella databastjänsterna uppvisar Azure SQL Database initialt en låg latens som sedan ökar proportionellt med datamängden, medan Google Cloud SQL visar en något högre latens vid lägre datamängder men mer konstant latens vid högre datamängder. Azure SQL visar sig vara mer kostnadseffektiv i förhållande till Google Cloud SQL, vilket gör den till ett mer fördelaktigt alternativ för företag som eftersträvar hög prestanda till lägre kostnader. Vid jämförelse mellan de två icke-relationella databastjänsterna Azure Cosmos DB och Google Cloud Datastore uppvisar Azure Cosmos DB genomgående jämförelsevis lägre latens och överlägsen kostnadseffektivitet. Detta gör Azure Cosmos DB till en fördelaktig lösning för företag som prioriterar ekonomisk effektivitet i sin databashantering. / As data volumes grow and the demands for scalability and availability within cloud services increase, the need for studies on their performance and cost-effectiveness is emphasized. These analyses are crucial for optimizing services and providing businesses with valuable recommendations to make well-grounded decisions about cloud data storage. This thesis examines cost and performance for relational and non-relational data storage solutions implemented on Microsoft Azure and Google Cloud Platform. The tool Hyperfine is used to evaluate latency and the cloud services cost efficiency is calculated using this result as well as their monthly cost. The study's results regarding relational data storage indicate that Azure SQL Database initially exhibits low latency, which then increases proportionally with the data volume, while Google Cloud SQL shows slightly higher latency at smaller data volumes but more consistent latency with more data. Azure SQL Database is more cost-effective, making it a more favorable option than Google Cloud SQL for companies seeking high performance at lower costs. Regarding evaluated services for non-relational data storage Azure Cosmos DB consistently demonstrates lower latency and superior cost efficiency compared to Google Cloud Datastore, making it the preferred solution for companies prioritizing economic efficiency in their database management.