Real-time filesystems. Guaranteeing timing constraints for disk accesses in RT-Mach

Abstract

Traditional real-time systems have largely avoided the use of disks due to their relative slow speeds and their unpredictability. However, many real-time upplications including multimedia systems and real-time database ap. plications benefit significantly from the use of disks to store and access real-time data. In this paper, we investigate the problem of obtaining guaranreed timely access to files on a disk in a real-time system. Our study focuses on several aspects of this problem of providing a real-time filesystem.First, we consider the use of two real-time disk scheduling algorithms: earliest deadline scheduling and just-in-time scheduling, a variation of aperiodic servers for ?he disk. The latter algorithm is designed to improve disk throughput that can be hurt when a real-time scheduling algorithm such as EDF is applied directly. Admission control policies with practically acceptable properties of performance and usabilio are provided. Next, we design and implement a real-time jilesystem on the RT-Mach microkernel-based system running a real-time shell. The new inteqace we develop is based on RT-Mach 's resource reservation paradigm and provides guaranteed and timelv access for multiple concurrent applications requiring disk bandwidth with direrent timing and volume requirements. Finally, we perform a detailed per$ormance evaluation of the real-time filesystem including its raw petjormance. We show the following positive but rather surprising result: our realrime scheduling filesystem not only provides guaranreed and timely access but also does so at relatively high levels of throughput. Traditional disk scheduling algorithms ofer completely unacceptable file access latencies for real-time applications and do so only at slightly higher throughput.

Empirical evaluation of latency-sensitive application performance in the cloud

“…Several techniques have been proposed for predictable allocation of the processor [14,15,19,25], network bandwidth [7,11,16] and disk throughput [2,23,18] Broadly speaking, these efforts have been targeted at adding quality-of-service (QoS) to operating systems. Several systems have been developed to integrate many resource allocation schemes into a single cohesive QoS-aware system.…”

Modular software architecture for flexible reservation mechanisms on heterogeneous resources

“…4 Real time disk schedulers usually assume to know a deadline per each request, or assign them one, and use modified versions of Earliest Deadline First (EDF [32]) to schedule requests. As an example, Molano et al in [33] propose JIT (Just In Time slack stealing), consisting in an EDF scheduler, modified in order to anticipate the service of requests near to the disk head position, postponing requests with smaller deadlines but with enough slack time to allow their completion even after being reordered. SCAN-EDF, introduced by Reddy et al in [42], reorders requests with the same deadline, sorting them in SCAN order.…”

High Throughput Disk Scheduling with Fair Bandwidth Distribution

“…In addition, examples of frameworks for providing QoS guarantees are Cello [20], APEX [21], PRISM [22] and Argon [23]. It is worth mentioning also real-time operating systems such as the Dresden Real-Time Operating System (DROPS) [5] and RT-Mach [6], and Real-Time Database Systems (RTDBS), which are architectures for performing database operations with real-time constraints [24].…”

“…Defined as slack of a request the difference between the deadline of the request and the earliest time by which the request can be served, Just-In-Time Slack Stealing (JIT) [6] is based on serving requests closer to the disk head instead of requests with lower deadlines but with large enough positive slack. SCAN-EDF [4] serves requests in EDF order, but if several requests have the same deadline, they are scheduled using a seek optimization algorithm (e.g., SCAN or C-LOOK).…”

“…In this respect, the problem of algorithms aimed only at maximising the disk throughput is that in the worst case they may delay the service of a request until the whole disk has been read or written (age-based policies, as in AS, can be added to mitigate the problem). In contrast, many schedulers, as e.g., SCAN-EDF [4], SATF-DAS [5], JIT [6], Hybrid [7], YFQ [8], pClock [9], adaptive DRR [10], CFQ [11] and adaptations of SFQ [12] have been proposed to provide control on request completion time and/or on bandwidth distribution, while at the same time trying to keep the throughput high.…”