Although the scalability of a distributed system is higher than a centralized system, big data solutions based on a distributed system may not provide better performance due to slower technological improvements in communication technologies compared to rapid advancements in CPU technologies. Also, low prices of storage units led to a centralized system being able to handle the entire data analysis workload of a mid-range enterprise. For example, 3 TB of data can be stored statically on a single computer and with the help of external drives and high performance computing tools that are connected to the network, a scalable centralized solution is achievable.
In this project, we will implement an analytics database using multicore CPUs and GPUs. Excellent knowledge of C++ is required.
About Project Supervisors
Kamer Kaya
kaya@sabanciuniv.edu