Ensuring performance isolation and differentiation among workloads that share a storage infrastructure is a basic requirement in consolidated data centers. Existing management tools rely on resource provisioning to meet performance goals; they require detailed knowledge of the system characteristics and the workloads. Provisioning is inherently slow to react to system and workload dynamics and, in the general case, it is not practical to provision for the worst case.We propose a software-only solution that ensures predictable performance for storage access. It is applicable to a wide range of storage systems and makes no assumptions about workload characteristics. We use an online feedback loop with an adaptive controller that throttles storage access requests to ensure that the available system throughput is shared among workloads according to their performance goals and their relative importance. The controller considers the system as a “black box” and adapts automatically to system and workload changes. The controller is distributed to ensure high availability under overload conditions, and it can be used for both block and file access protocols. The evaluation of Triage , our experimental prototype, demonstrates workload isolation and differentiation in an overloaded cluster file-system where workloads and system components are changing.

Data replication is used extensively in wide-area distributed systems to achieve low data-access latency. A large number of heuristics have been proposed to perform replica placement. Practical experience indicates that the choice of heuristic makes a big difference in terms of the cost of required infrastructure (e.g., storage capacity and network bandwidth), depending on system topology, workload and performance goals. We describe a method to assist system designers choose placement heuristics that meet their performance goals for the lowest possible infrastructure cost. Existing heuristics are classified according to a number of properties. The inherent cost (lower bound) for each class of heuristics is obtained for given system, workload and performance goals. The system designer compares different classes of heuristics on the basis of these lower bounds. Experimental results show that choosing a heuristic with the proposed methodology results in up to 7 times lower cost compared to using an "obvious " heuristic, such as caching.

Ensuring performance isolation and differentiation among workloads that share a storage infrastructure is a basic requirement in consolidated data centers. Existing management tools rely on resource provisioning to meet performance goals; they require detailed knowledge of the system characteristics and the workloads. Provisioning is inherently slow to react to system and workload dynamics, and in the general case, it is impossible to provision for the worst case. We propose a software-only solution that ensures predictable performance for storage access. It is applicable to a wide range of storage systems and makes no assumptions about workload characteristics. We use an on-line feedback loop with an adaptive controller that throttles storage access requests to ensure that the available system throughput is shared among workloads according to their performance goals and their relative importance. The controller considers the system as a "black box" and adapts automatically to system and workload changes. The controller is distributed to ensure high availability under overload conditions, and it can be used for both block and file access protocols. The evaluation of Triage, our experimental prototype, demonstrates workload isolation and differentiation, in an overloaded cluster file-system where workloads and system components are changing.

Distributed hash table (DHT)-based overlay networks, represented by Pastry, CAN, and Chord, offer an administration-free and fault-tolerant application-level overlay network. While elegant from a theoretical perspective, these systems have some disadvantages. First, they rely on application-level routing, which may be inefficient with respect to network delays and bandwidth consumption. Second, they typically construct a homogeneously structured overlay even though nodes in these networks usually have varying physical connectivity and packet-forwarding capacities. In this paper, we propose two approaches for constructing an auxiliary expressway network to take advantage of the different connectivity, forwarding capacities, and availabilities of the nodes. As a result, we are able to reconcile the conflict of presenting the applications with a homogeneous structured overlay to simplify management, while at the same time taking advantage of the inherent heterogeneity of the underlying physical network to speed up routing. Our simulation results show that our expressway can achieve close to optimal routing performance (on average, 1.07 and 1.41 times optimal routing for an Internet-like topology and a large synthesized transit-stub graph, respectively) in overlay networks.

Our goal is to develop a Web server log analysis tool that produces a Web site profile and its system resource usage in a way useful to service providers. Understanding the nature of traffic to the Web site is crucial in properly designing site support infrastructure, especially for large, busy sites. The main questions we address are the new access patterns of today's WWW, how to characterize dynamics or evolution of Web sites, and how to measure the rate of changes. We propose a set of new metrics to characterize the site dynamics, and we illustrate them with analysis of three different Web sites.

Prefetching offers the potential to improve the performance of linked data structure (LDS) traversals. However, previously proposed prefetching methods only work well when there is enough work processing a node that the prefetch latency can be hidden, or when the LDS is long enough and the traversal path is known a priori. This paper presents a prefetching technique called prefetch arrays which can prefetch both short LDS, as the lists found in hash tables, and trees when the traversal path is nor known a priori. We offer two implementations, one software-only and one which combines software annotations with a prefetch engine in hardware. On a pointer-intensive benchmark suite, we show that our implementations reduce the memory stall lime by 23% to 51% for the kernels with linked lists, while the other prefetching methods cause reductions that are substantially less. For binary-trees, our hardware method manages to cut nearly 60% of the memory stall time even when the traversal path is not known a priori. However, when the branching factor of the tree is too high, our technique does not improve performance. Another contribution of the paper is that we quantify pointer-chasing found in interesting applications such as OLTP, Expert Systems, DSS, and JAVA codes and discuss which prefetching techniques are relevant to use in each case.

This paper presents an analytical model to study how working sets scale with database size and other applications parameters in decision-support systems (DSS). The model uses application parameters, that are measured on down-scaled database executions, to predict cache miss ratios for executions of large databases.

We consider a network of workstations (NOW) organization consisting of a number of bus-based multiprocessor servers interconnected by an ATM switch. A shared-memory model is supported by distributed virtual shared memory (DVSM) and this paper focuses on the access penalties incurred by (1) ATM and (2) the DVSM software. First, through detailed architectural simulations we find that while the bandwidth and the latency of the ATM switch fabrics are found to be acceptable, the latency incurred by commercially available ATM interfaces has a first order effect on the performance. We also study the effects of various scheduling policies for the coherence handlers. Our data suggest that since the probability of finding an idle processor within a cluster is high, a good policy is to schedule it there instead of letting an extra compute processor execute coherence handlers. Overall, by adjusting the adaptation layer of ATM to a DVSM system we find that ATM is a promising technology for these kinds of systems.