John Ardelius

On the effects of caching in access aggregation networks

All forecasts of Internet traffic point at a substantial growth over the next few years. From a network operator perspective, efficient in-network caching of data is and will be a key component in trying to cope with and profit from this increasing demand. One problem, however, is to evaluate the performance of different caching policies as the number of available data items as well as the distribution networks grows very large. In this work, we develop an analytical model of an aggregation access network receiving a continuous flow of requests from external clients. We provide exact analytical solutions for cache hit rates, data availability and more. This enables us to provide guidelines and rules of thumb for operators and Information-Centric Network designers. Finally, we apply our analytical results to a real VoD trace from a network operator and show that substantial bandwidth savings can be expected when using in-network caching in a realistic setting.

Circumspect descent prevails in solving random constraint satisfaction problems

We study the performance of stochastic local search algorithms for random instances of the K-satisfiability (K-SAT) problem. We present a stochastic local search algorithm, ChainSAT, which moves in the energy landscape of a problem instance by never going upwards in energy. ChainSAT is a focused algorithm in the sense that it focuses on variables occurring in unsatisfied clauses. We show by extensive numerical investigations that ChainSAT and other focused algorithms solve large K-SAT instances almost surely in linear time, up to high clause-to-variable ratios α; for example, for K = 4 we observe linear-time performance well beyond the recently postulated clustering and condensation transitions in the solution space. The performance of ChainSAT is a surprise given that by design the algorithm gets trapped into the first local energy minimum it encounters, yet no such minima are encountered. We also study the geometry of the solution space as accessed by stochastic local search algorithms.

Predicting real-time service-level metrics from device statistics

While real-time service assurance is critical for emerging telecom cloud services, understanding and predicting performance metrics for such services is hard. In this paper, we pursue an approach based upon statistical learning whereby the behavior of the target system is learned from observations. We use methods that learn from device statistics and predict metrics for services running on these devices. Specifically, we collect statistics from a Linux kernel of a server machine and predict client-side metrics for a video-streaming service (VLC). The fact that we collect thousands of kernel variables, while omitting service instrumentation, makes our approach service-independent and unique. While our current lab configuration is simple, our results, gained through extensive experimentation, prove the feasibility of accurately predicting client-side metrics, such as video frame rates and RTP packet rates, often within 10–15% error (NMAE), also under high computational load and across traces from different scenarios.

Behavior of heuristics on large and hard satisfiability problems.

We study the behavior of ASAT, a heuristic for solving satisfiability problems by stochastic local search near the SAT/UNSAT transition. The heuristic is focused, i.e. only variables in unsatisfied clauses are updated in each step, and is significantly simpler, while similar to, walksat or Focused Metropolis Search. We show that ASAT solves instances as large as N = 10 in linear time, on average, up to α = 4.21 for random 3SAT. For K higher than 3, ASAT appears to solve instances at the Montanari-Ricci-Tersenghi-Parisi “FRSB threshold” αs(K) in linear time, up to K = 7.

Predicting service metrics for cluster-based services using real-time analytics

Predicting the performance of cloud services is intrinsically hard. In this work, we pursue an approach based upon statistical learning, whereby the behaviour of a system is learned from observations. Specifically, our testbed implementation collects device statistics from a server cluster and uses a regression method that accurately predicts, in real-time, client-side service metrics for a video streaming service running on the cluster. The method is service-agnostic in the sense that it takes as input operating-systems statistics instead of service-level metrics. We show that feature set reduction significantly improves prediction accuracy in our case, while simultaneously reducing model computation time. We also discuss design and implementation of a real-time analytics engine, which processes streams of device statistics and service metrics from testbed sensors and produces model predictions through online learning.

Exhaustive enumeration unveils clustering and freezing in the random 3-satisfiability problem.

We study geometrical properties of the complete set of solutions of the random 3-satisfiability problem. We show that even for moderate system sizes the number of clusters corresponds surprisingly well with the theoretic asymptotic prediction. We locate the freezing transition in the space of solutions, which has been conjectured to be relevant in explaining the onset of computational hardness in random constraint satisfaction problems.

Manycore work stealing

This paper investigates executing task based programs on a 64 core Tilera processor under the high performance work stealer Wool. We measure the performance of several programs from the BOTS benchmark suite, observing excellent scalability whenever sufficient parallelism exists. We also explore alternatives to random victim selection; we use sampling to try to find a large task to steal and set based stealing to improve cache and TLB locality.

Clustering of solutions in hard satisfiability problems

We study numerically the solution space structure of random 3-SAT problems close to the SAT/UNSAT transition. This is done by considering chains of satisfiability problems, where clauses are added ...

A platform for predicting real-time service-level metrics from device statistics

While real-time service assurance is critical for emerging telecom cloud services, understanding and predicting performance metrics for such services is hard. In this paper, we pursue an approach based upon statistical learning whereby the behavior of the target system is learned from observations. We use methods that learn from device statistics and predict metrics for services running on these devices. Specifically, we collect statistics from a Linux kernel of a server machine and predict client-side metrics for a video-streaming service (VLC). The fact that we collect thousands of kernel variables, while omitting service instrumentation, makes our approach service-independent and unique. While our current lab configuration is simple, our results, gained through extensive experimentation, prove the feasibility of accurately predicting client-side metrics, such as video frame rates and RTP packet rates, often within 10-15% error (NMAE), also under high computational load and across traces from different scenarios.

Brief announcement: the accuracy of tree-based counting in dynamic networks

We study a simple Bellman-Ford-like protocol which performs network size estimation over a tree-shaped overlay. A continuous time Markov model is constructed which allows key protocol characteristics to be estimated under churn, including the expected number of nodes at a given (perceived) distance to the root and, for each such node, the expected (perceived) size of the subnetwork rooted at that node. We validate the model by simulations, using a range of network sizes, node degrees, and churn-to-protocol rates, with convincing results.