Results 1 - 10
of
18
User-Level Communication in Cluster-Based Servers
- In Proceedings of the 8th IEEE International Symposium on High-Performance Computer Architecture (HPCA 8
, 2002
"... Clusters of commodity computers are currently being used to provide the scalability required by several popular Internet services. In this paper we evaluate an efficient cluster-based WWW server, as a function of the characteristicsof the intra-cluster communication architecture. More specifically, ..."
Abstract
-
Cited by 30 (11 self)
- Add to MetaCart
Clusters of commodity computers are currently being used to provide the scalability required by several popular Internet services. In this paper we evaluate an efficient cluster-based WWW server, as a function of the characteristicsof the intra-cluster communication architecture. More specifically, we evaluate the impact of processor overhead, networkbandwidth, remote memory writes, and zero-copy data transfers on the performance of our server. Our experimental results with an 8-node cluster and four real WWW traces show that networkbandwidth affects the performanceof our server by only 6%. In contrast, user-level communication can improve performance by as much as 29%. Low processor overhead, remote memory writes, and zero-copyall make small contributions towardsthis overall gain. Tobe able to extrapolate fromour experimental results, we usean analytical model to assess the performance of our server under different workload characteristics, different numbers of cluster nodes, and higher performance systems. Our modeling results show that higher gains (of up to 55%) can be accrued for workloads with large working sets and next-generation servers running on large clusters. 1
Performance Analysis of Remote File System Access over High Bandwidth Local Network
, 2003
"... We study the performance of file servers, comparing NFS implementation in Linux to our experimental lightweight system called ORFA. The aim is to find out NFS bottlenecks in the case of high bandwidth local network. Using a simple protocol without cache allow us to get best performance from the unde ..."
Abstract
-
Cited by 10 (8 self)
- Add to MetaCart
(Show Context)
We study the performance of file servers, comparing NFS implementation in Linux to our experimental lightweight system called ORFA. The aim is to find out NFS bottlenecks in the case of high bandwidth local network. Using a simple protocol without cache allow us to get best performance from the underlying communication subsystem. Our user-level implementation avoids several kernel-level constraints and thus provides better performance than NFS. Moreover we explore several optimization techniques to reduce the server overhead which usually is the bottleneck.
User-Centric Data Migration in Networked Storage Systems
"... This paper considers the problem of balancing locality and load in networked storage systems with multiple storage devices (or bricks). Data distribution affects locality and load balance across the devices in a networked storage system. This paper proposes a user-centric data migration scheme which ..."
Abstract
-
Cited by 7 (4 self)
- Add to MetaCart
(Show Context)
This paper considers the problem of balancing locality and load in networked storage systems with multiple storage devices (or bricks). Data distribution affects locality and load balance across the devices in a networked storage system. This paper proposes a user-centric data migration scheme which tries to balance locality and load in such networked storage systems. The presented approach automatically and transparently manages migration of data blocks among disks as data access patterns and loads change over time. We implemented a prototype system, embodying our ideas, on PCs running Linux. This paper presents the design of user-centric migration and an evaluation of it through realistic experiments. 1.
Linux NFS Client Write Performance
, 2001
"... We introduce a simple sequential write benchmark and use it to improve Linux NFS client write performance. We reduce the latency of the write() system call, improve SMP write performance, and reduce kernel CPU processing during sequential writes. Cached write throughput to NFS files improves by more ..."
Abstract
-
Cited by 7 (0 self)
- Add to MetaCart
We introduce a simple sequential write benchmark and use it to improve Linux NFS client write performance. We reduce the latency of the write() system call, improve SMP write performance, and reduce kernel CPU processing during sequential writes. Cached write throughput to NFS files improves by more than a factor of three.
T.D.: A Cost-Effective Distributed File Service with QoS Guarantees.
, 2007
"... Abstract. Large-scale, value-added Internet services composed of independent cooperating or competing services will soon become common place. Several groups have addressed the performance, communication, discovery, and description aspects of these services. However, little work has been done on eff ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
Abstract. Large-scale, value-added Internet services composed of independent cooperating or competing services will soon become common place. Several groups have addressed the performance, communication, discovery, and description aspects of these services. However, little work has been done on effectively composing paid services and the quality-of-service (QoS) guarantees that they provide. We address these issues in the context of distributed file storage in this paper. In particular, we propose, implement, and evaluate a cost-effective, QoSaware distributed file service comprising a front-end file service and back-end (third-party) storage services. Our front-end service uses mathematical modeling and optimization to provide performance and availability guarantees at low cost by carefully orchestrating the accesses to the back-end services. Experimental results from our prototype implementation validate our modeling and optimization. We conclude that our approach for providing QoS at low cost should be useful to future composite Internet services.
Prioritizing Write Acknowledgment inside Network Fileservers
"... Abstract—Output of network file servers exhibits bursty traffic patterns, and this sometimes contends with control traffic. An example is contention between data traffic and write acknowl-edgments on a loaded server. In such a situation, we can improve the write performance by prioritizing write ack ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract—Output of network file servers exhibits bursty traffic patterns, and this sometimes contends with control traffic. An example is contention between data traffic and write acknowl-edgments on a loaded server. In such a situation, we can improve the write performance by prioritizing write acknowledgments at the network interface of the server. To validate this scheme, we conducted an empirical study of the prioritization of write acknowledgments. Systematic experiments revealed that the pro-posed scheme can improve write latency and throughput of loaded servers without influencing read performance. The results suggested that the technique is widely applicable to systems where transactions are a mixture of read and write requests, and especially if the degree of concurrency is high. I.
Major Subject: Computer EngineeringFLEXIBLE ALLOCATION AND SPACE MANAGEMENT IN STORAGE SYSTEMS
, 2007
"... In this dissertation, we examine some of the challenges faced by the emerging networked storage systems. We focus on two main issues. Current file systems allocate storage statically at the time of their creation. This results in many suboptimal scenarios, for example: (a) space on the disk is not a ..."
Abstract
- Add to MetaCart
(Show Context)
In this dissertation, we examine some of the challenges faced by the emerging networked storage systems. We focus on two main issues. Current file systems allocate storage statically at the time of their creation. This results in many suboptimal scenarios, for example: (a) space on the disk is not allocated well across multiple file systems, (b) data is not organized well for typical access patterns. We propose Virtual Allocation for flexible storage allocation. Virtual allocation separates storage allocation from the file system. It employs an allocate-on-write strategy, which lets applications fit into the actual usage of storage space without regard to the configured file system size. This improves flexibility by allowing storage space to be shared across different file systems. We present the design of virtual allocation and an evaluation of it through benchmarks based on a prototype system on Linux. Next, based on virtual allocation, we consider the problem of balancing locality and load in networked storage systems with multiple storage devices (or bricks). Data distribution affects locality and load balance across the devices in a networked storage system. We propose user-optimal data migration scheme which tries to balance
1 Managing Storage Space in a Flash and Disk Hybrid Storage System
"... This paper considers the problem of efficiently managing storage space in a hybrid storage system employing flash and disk drives. The flash and disk drives exhibit different performance characteristics of read and write behavior. We propose a technique for balancing the workload properties across f ..."
Abstract
- Add to MetaCart
(Show Context)
This paper considers the problem of efficiently managing storage space in a hybrid storage system employing flash and disk drives. The flash and disk drives exhibit different performance characteristics of read and write behavior. We propose a technique for balancing the workload properties across flash and disk drives in such a hybrid storage system. We consider various alternatives for managing the storage space in such a hybrid system and show that the proposed technique improves performance in diverse scenarios. The presented approach automatically and transparently manages migration of data blocks among flash and disk drives based on their access patterns. We implemented a prototype storage system employing our ideas. This paper presents the design and an evaluation of the proposed approach through realistic experiments. I.
Managing Energy Usage and Cost Through Load Distribution in Multi-Data-center Services
, 2013
"... Multi-data-center services will soon be common place. These services raise many questions about how exactly to distribute the offered load across the data centers. In fact, different load distributions may produce wildly different monetary costs, performance, and/or energy consumption. Moreover, the ..."
Abstract
- Add to MetaCart
(Show Context)
Multi-data-center services will soon be common place. These services raise many questions about how exactly to distribute the offered load across the data centers. In fact, different load distributions may produce wildly different monetary costs, performance, and/or energy consumption. Moreover, these large services consume large amounts of energy produced via carbon-dioxide-intensive means. We refer to this as “brown energy”, in contrast to renewable or “green ” energy. This large energy consumption represents significant and fast-growing financial and environmental costs. Increasingly, services are exploring dynamic methods to manage energy and energy-related costs while respecting their service-level agreements (SLAs). Furthermore, it will soon be important for these services to manage their usage of brown or green energy. Based on these observations, this dissertation explores load distribution policies that minimize cost or energy consumption, while respecting the services’ performance requirements. First, we design, implement, and evaluate software support for multidata-center
Performance Analysis of Remote File System Access over High Bandwidth Local Network
, 2003
"... We study the performance of file servers, comparing NFS implementation in Linux to our experimental lightweight system called ORFA. The aim is to find out NFS bottlenecks in the case of high bandwidth local network. ..."
Abstract
- Add to MetaCart
(Show Context)
We study the performance of file servers, comparing NFS implementation in Linux to our experimental lightweight system called ORFA. The aim is to find out NFS bottlenecks in the case of high bandwidth local network.
