Over a few years back, REST APIs were considered standard web APIs, which nowhave a strong competitor. REST APIs provide some excellent features like stateless serversand structured access to resources. However, over time, it doesn’t offer flexibility withthe access of data and client changing requirements. In 2015 GraphQL was introduced byFacebook, which overcomes the problems with the REST and provides more flexibility andefficiency to the client requirements. For example, remove the over and under fetching.To change the existing APIs into GraphQL APIs require considerable time and effort.Therefore, some server implementation tools are developed to reduce the developmentcost and time. A few of these tools generate GraphQL schema and server implementationsautomatically over a legacy database.This master thesis studies tools that automatically generate GraphQL server implementationover legacy databases and evaluate such generated GraphQL server’s performance.First, we find some GraphQL server implementation tools such as Hasura andPostGraphile and then compare the server’s performance using benchmark methodology.Secondly, we run an experiment on a computer system and use the performance metricsfor assessment. The results of our experiment concluded that PostGraphile has higherthroughput and low query execution time as compared to Hasura. In most of the querytemplates from the benchmark, PostGraphile outperforms Hasura.

Several cloud providers offer reltional databases as part of their portfolio. It is however not obvious how resource virtualization and sharing, which is inherent to cloud computing, influence performance and predictability of these cloud databases. Cloud providers give little to no guarantees for consistent execution or isolation from other users. To evaluate the performance isolation capabilities of two commercial cloud databases, we ran a series of experiments over the course of a week (a query, a minute) and report variations in query response times. As a baseline, we ran the same experiments on a dedicated server in our data center. The results show that in the cloud single outliers are up to 31 times slower than the average. Additionally, one can see a point in time after which the average performance of all executed queries improves by 38 %.

As databases and traffic over the internet is becoming larger by the day, the performance of sending information has become a target of great importance. In past years, other software architectural styles such as REST have been used as it is a reliable framework and works really well when one has a dependable internet connection. In 2015, the querying language GraphQL was released by Facebook to the public as an alternative to REST. GraphQL made improvements in fetching data by for example removing the possibility of under- and overfitting. This means that a client only gets the data which they have requested, nothing more, nothing less. To create a GraphQL schema and server implementation requires time, effort and knowledge. This is however a requirement to run GraphQL over your current legacy database. For this reason multiple server implementation tools have been created by vendors to reduce development time and instead auto-generates a GraphQL schema and server implementation using an already existing database. This bachelor thesis will pick, run and compare the benchmarks of the two different server implementation tools Hasura and PostGraphile. This is done using a benchmark methodology based on technical difficulties (choke points). The result of our benchmark suggests that the throughput is larger for Hasura compared to PostGraphile whilst the query execution time as well as query response time is similar. PostGraphile is better at paging without offset as well as ordering, but on all other cases Hasura outperforms PostGraphile or shows similar results. / Linköping GraphQL Benchmark (LinGBM)