Results 1 - 10
of
19
Hippodrome: Running Circles around Storage Administration
- In Proceedings of the Conference on File and Storage Technologies
, 2002
"... Enterprise-scale computer storage systems are extremely difficult to manage due to their size and complexity. It is difficult to generate a good storage system design for a given workload and to correctly implement the selected design. Traditionally, initial system configuration is performed by admi ..."
Abstract
-
Cited by 142 (10 self)
- Add to MetaCart
(Show Context)
Enterprise-scale computer storage systems are extremely difficult to manage due to their size and complexity. It is difficult to generate a good storage system design for a given workload and to correctly implement the selected design. Traditionally, initial system configuration is performed by administrators who are guided by rules of thumb. Unfortunately, this process involves trial and error, and as a result is tedious and error-prone. In this paper, we introduce Hippodrome, an approach to automating initial system configuration. Hippodrome is an iterative loop that analyzes an existing system to determine its requirements, creates a new storage system design to better meet these requirements, and migrates the existing system to the new design. In this paper, we show how Hippodrome automates initial system configuration. 1
An Admission Control Scheme for Predictable Server Response Time for Web Accesses
- In Proceedings of the 10th World Wide Web Conference, Hong Kong
, 2001
"... The diversity in web object types and their resource requirements contributes to the unpredictability of web service provisioning. In this paper, an efficient admission control algorithm, PACERS, is proposed to provide different levels of services based on the server workload characteristics. Servic ..."
Abstract
-
Cited by 61 (4 self)
- Add to MetaCart
(Show Context)
The diversity in web object types and their resource requirements contributes to the unpredictability of web service provisioning. In this paper, an efficient admission control algorithm, PACERS, is proposed to provide different levels of services based on the server workload characteristics. Service quality is ensured by periodical allocation of system resources based on the estimation of request rate and service requirements of prioritized tasks. Admission of lower priority tasks is restricted during high load periods to prevent denial-of-services to high priority tasks. A double-queue structure is implemented to reduce the effects of estimation inaccuracy and to utilize the spare capacity of the server, thus increasing the system throughput. Response delays of the high priority tasks are bounded by the length of the prediction period. Theoretical analysis and experimental study show that the PACERS algorithm provides desirable throughput and bounded response delay to the prioritized tasks, without any significant impact on the aggregate throughput of the system under various workload.
Chameleon: a self-evolving, fully-adaptive resource arbitrator for storage systems
- In Proceedings of the 2005 USENIX Technical Conference
, 2005
"... Enterprise applications typically depend on guaranteed performance from the storage subsystem, lest they fail. However, unregulated competition is unlikely to result in a fair, predictable apportioning of resources. Given that widespread access protocols and scheduling policies are largely best-effo ..."
Abstract
-
Cited by 22 (2 self)
- Add to MetaCart
(Show Context)
Enterprise applications typically depend on guaranteed performance from the storage subsystem, lest they fail. However, unregulated competition is unlikely to result in a fair, predictable apportioning of resources. Given that widespread access protocols and scheduling policies are largely best-effort, the problem of providing performance guarantees on a shared system is a very difficult one. Clients typically lack accurate information on the stor-age system’s capabilities and on the access patterns of the workloads using it, thereby compounding the problem. CHAMELEON is an adaptive arbitrator for shared storage resources; it relies on a combination of self-refining mod-els and constrained optimization to provide performance guarantees to clients. This process depends on minimal information from clients, and is fully adaptive; decisions are based on device and workload models automatically in-ferred, and continuously refined, at run-time. Corrective actions taken by CHAMELEON are only as radical as war-ranted by the current degree of knowledge about the sys-tem’s behavior. In our experiments on a real storage sys-tem CHAMELEON identified, analyzed, and corrected per-formance violations in 3-14 minutes—which compares very favorably with the time a human administrator would have needed. Our learning-based paradigm is a most promis-ing way of deploying large-scale storage systems that ser-vice variable workloads on an ever-changing mix of device types. 1
ACES: An Efficient Admission Control Scheme for QoS-Aware Web Servers
- Computer Communications
, 2003
"... The unpredictability of server response performance hinders the advance of new application on the Internet. In this paper, we present an efficient admission control algorithm, ACES, based on the server workload characteristics. The admission control algorithm ensures the bounded response time fro ..."
Abstract
-
Cited by 12 (0 self)
- Add to MetaCart
(Show Context)
The unpredictability of server response performance hinders the advance of new application on the Internet. In this paper, we present an efficient admission control algorithm, ACES, based on the server workload characteristics. The admission control algorithm ensures the bounded response time from a web server by periodical allocation of system resources according to the resource requirements of incoming tasks. By rejecting requests exceeding server capacity, the response performance of the server is well maintained even under high system utilization. The resource requirements of tasks are estimated based on their types. A double-queue structure is implemented to reduce the effects caused by estimation inaccuracy, and to exploit the spare capacity of the server, thus increasing the system throughput. The admission control algorithm can be used for server overload control and for QoS provisioning of service differentiating Internet servers. Response delays of accepted tasks are bounded by the desired predefined time period.
Efficient data migration in self-managing storage systems
- In Proc. ICAC ’06
, 2006
"... Abstract — Self-managing storage systems automate the tasks of detecting hotspots and triggering data migration to alleviate them. This paper argues that existing data migration techniques do not minimize data copying overhead incurred during reconfiguration, which in turn impacts application perfor ..."
Abstract
-
Cited by 9 (3 self)
- Add to MetaCart
(Show Context)
Abstract — Self-managing storage systems automate the tasks of detecting hotspots and triggering data migration to alleviate them. This paper argues that existing data migration techniques do not minimize data copying overhead incurred during reconfiguration, which in turn impacts application performance. We propose a novel technique that automatically detects hotspots and uses the bandwidth-to-space ratio metric to greedily reconfigure the system while minimizing the resulting data copying overhead. We validate our technique with simulations and a prototype implemented into the Linux Kernel. Our prototype and simulations show our algorithm successfully eliminates hotspots with a factor of two reduction in data copying overhead compared to other approaches. I.
Fault-tolerant replication management in large-scale distributed storage systems
- Proceedings 18th IEEE Symposium on Reliable Distributed Systems
, 1999
"... ..."
(Show Context)
Virtualizing Disk Performance
, 2008
"... Large- and small-scale storage systems frequently serve a mixture of workloads, an increasing number of which require some form of performance guarantee. Providing guaranteed disk performance—the equivalent of a “virtual disk”—is challenging because disk requests are nonpreemptible and their executi ..."
Abstract
-
Cited by 8 (4 self)
- Add to MetaCart
(Show Context)
Large- and small-scale storage systems frequently serve a mixture of workloads, an increasing number of which require some form of performance guarantee. Providing guaranteed disk performance—the equivalent of a “virtual disk”—is challenging because disk requests are nonpreemptible and their execution times are stateful, partially non-deterministic, and can vary by orders of magnitude. Guaranteeing throughput, the standard measure of disk performance, requires worst-case I/O time assumptions orders of magnitude greater than average I/O times, with correspondingly low performance and poor control of the resource allocation. We show that disk time utilization— analogous to CPU utilization in CPU scheduling and the only fully provisionable aspect of disk performance—yields greater control, more efficient use of disk resources, and better isolation between request streams than bandwidth or I/O rate when used as the basis for disk reservation and scheduling.
SmartMig: Risk-modulated Proactive Data Migration for Maximizing Storage System Utility
- In Proc. of IEEE MSST
, 2006
"... The goal of storage management is to maximize the overall utility of the storage system by continuously tuning the amount of resources allocated to multiple independent competing applications. Due to variations in access characteristics, service level objectives, and exception events such as failure ..."
Abstract
-
Cited by 7 (0 self)
- Add to MetaCart
(Show Context)
The goal of storage management is to maximize the overall utility of the storage system by continuously tuning the amount of resources allocated to multiple independent competing applications. Due to variations in access characteristics, service level objectives, and exception events such as failures and load surges, there is a need to invoke corrective actions such as data migration to modify the resources allocated to a given application. There is a significant body of research on automated data migration – their focus has primarily been on optimizing for the current system load without considering load forecasts; the scheduling of the migration operation today is currently heuristic and coarse-grained; finally, there is a need to factor in prediction inaccuracies and migration data-size (referred to as risks) in the decision-making. This paper proposes SMARTMIG: a framework for optimizing the storage utility by proactively scheduling data migration using time-series forecasts. SMARTMIG generates several plans for what data to migrate, where to migrate, how to migrate (i.e., the migration speed), and when to migrate. These plans are generated using constraint optimization, and their selection is modulated by risk analysis of the prediction accuracy and the migration overheads. For the experimental evaluation of SMARTMIG, we developed a detailed storage system simulator, and analyzed the quality of migration decisions made in different scenarios. Our results show that for a significant percentage of scenarios, SMARTMIG in a automated fashion minimizes the utility loss by 80 % compared to no action invocation. 1.
Disk array models in Minerva
, 2001
"... storage systems, disk arrays, analytical models Enterprise storage systems typically depend on disk arrays to satisfy their capacity and availability needs. To design and maintain storage systems that efficiently satisfy evolving requirements, it is critical to be able to evaluate configuration alte ..."
Abstract
-
Cited by 7 (2 self)
- Add to MetaCart
(Show Context)
storage systems, disk arrays, analytical models Enterprise storage systems typically depend on disk arrays to satisfy their capacity and availability needs. To design and maintain storage systems that efficiently satisfy evolving requirements, it is critical to be able to evaluate configuration alternatives without having to physically implement them. Because of the large number of candidate configurations that need to be evaluated in real-life situations, simulation models are excessively slow for that task. In this paper, we describe analytical throughput models for RAID 1/0 and RAID 5 storage in the Hewlett-Packard FC-30 disk array. We validate our models against the real array, and report the relative errors in the models ’ predictions. Our models have a mean error of 5.4 % and a maximum error of 19%, for the set of validations workloads we used.
Researching System Administration
, 2002
"... System administration is a phenomenally important, yet surprisingly ignored sub-field of Computer Science. We hypothesize that this avoidance is because approaches for performing academic research on system administration problems are not well known. To reduce the difficulty of performing research ..."
Abstract
-
Cited by 6 (0 self)
- Add to MetaCart
System administration is a phenomenally important, yet surprisingly ignored sub-field of Computer Science. We hypothesize that this avoidance is because approaches for performing academic research on system administration problems are not well known. To reduce the difficulty of performing research, we present a small set of principles that can be used to evaluate solutions, a classification of existing research on system administration, and three approaches to research on system administration that we illustrate with the research that we have done.
