Important stakeholders in the personal-OS arena have switched to the BFQ I/O scheduler, to endow their systems with a very low I/O latency. The story differs with enterprise systems tough. By controlling I/O with BFQ, these systems would reach a 5X-10X throughput boost. Nevertheless, almost all stakeholders still stick to uncontrolled I/O. For various reasons they fail to see the economic benefits that may follow from the above boost.
At the other end of the spectrum, Facebook seems to see these benefits so clearly that they have invested in the design and implementation of two brand new I/O controllers for Linux. The latter control, respectively, latency and bandwidth. And these are the two main goals of BFQ.
In this presentation we try to shed a light on this sort of controversial situation, by showing some performance figures. In
particular we show again (the poor) performance without I/O control, after which we compare the performance of these new controllers with that of BFQ.
Assistant Professor (Linaro)
Paolo Valente is an Assistant Professor of Computer Science at the University of Modena and Reggio Emilia, Italy, and a collaborator of the Linaro engineering organization. Paolos main activities focus on scheduling algorithms for storage devices, transmission links and CPUs. In this respect, Paolo is the author of the last version of the BFQ I/O scheduler. BFQ entered the Linux kernel from 4.12, providing unprecedented low-latency and fairness guarantees. As for transmission links, Paolo is one of the authors of the QFQ packet scheduler, which has been in the Linux kernel until 3.7, after that it has been replaced by QFQ+, a faster variant defined and implemented by Paolo himself. Finally, Paolo has also defined and implemented other algorithms, part of which are now in FreeBSD, and has provided new theoretic results on multiprocessor scheduling.