Deshi Ye

Virt-LM: a benchmark for live migration of virtual machine

Virtualization technology has been widely applied in data centers and IT infrastructures, with advantages of server consolidation and live migration. Through live migration, data centers could flexibly move virtual machines among different physical machines to balance workloads, reduce energy consumption and enhance service availability. Todays data centers can grow to a huge scale. This implies that frequent live migration would be desirable for the economic use of hardware resources. Then, the performance of the live migration strategy will be an issue. So, we need a reliant evaluation method to choose the software and hardware environments that will produce the best live migration performance. However, there is not a complete live migration benchmark available currently. In addition, the existing evaluation methodologies select different metrics, different workloads and different test means. Thus, it is difficult to compare their results. In this paper we first survey the current research and their evaluation methods on live migration. We then summarize the critical issues for the live migration evaluation and also raise other unreported potential problems. We propose our solutions and present an implementation in our live migration benchmark -- Virt-LM. This is a benchmark for comparing live migration performance among different software and hardware environments in a data center scenario. We detail its design and provide some experimental results to validate its effectiveness.

A note on online strip packing

Abstract In online strip packing we are asked to pack a list of rectangles one by one into a vertical strip of unit width, without any information about future rectangles. The goal is to minimize the total height of strip used. The best known algorithm is First Fit Shelf algorithm (Baker and Schwarz in SIAM J. Comput. 12(3):508–525, 1983), which has an absolute competitive ratio of 6.99 under the assumption that the height of each rectangle is bounded from above by one. We improve the shelf algorithm and show an absolute competitive ratio of

Online multiple-strip packing

7/2+\sqrt{10}\approx 6.6623

On-Line Scheduling of Parallel Jobs

without the restriction on rectangle heights. Our algorithm also beats the best known online algorithm for parallel job scheduling.

Frequency allocation problems for linear cellular networks

We study an online multiple-strip packing problem, whose goal is to pack the given rectangles into m vertical strips of unit width such that the maximum height used among the strips is minimized. Rectangles arrive one by one. The decision of delivering the rectangles to a strip as well as packing them into the strip must be done immediately and irrevocably without any information on the next rectangles. Both randomized and deterministic online algorithms are investigated, all of which are guaranteed a constant competitive ratio.

Two Optimization Mechanisms to Improve the Isolation Property of Server Consolidation in Virtualized Multi-core Server

We study an on-line parallel job scheduling problem, where jobs arrive over list. A parallel job may require a number of machines for its processing at the same time. Upon arrival of a job, its processing time and the number of requested machines become known, and it must be scheduled immediately without any knowledge of future jobs. We present a 8-competitive on-line algorithm, which improves the previous upper bound of 12 by Johannes [8]. Furthermore, we investigate two special cases in which jobs arrive in non-increasing order of processing times or jobs arrive in non-increasing order of sizes. Better bounds are shown.

Online frequency allocation in cellular networks

We study the online frequency allocation problem for wireless linear (highway) cellular networks, where the geographical coverage area is divided into cells aligned in a line. Calls arrive over time and are served by assigning frequencies to them, and no two calls emanating from the same cell or neighboring cells are assigned the same frequency. The objective is to minimize the span of frequencies used. In this paper we consider the problem with or without the assumption that calls have infinite duration. If there is the assumption, we propose an algorithm with absolute competitive ratio of 3/2 and asymptotic competitive ratio of 1.382. The lower bounds are also given: the absolute one is 3/2 and the asymptotic one is 4/3. Thus, our algorithm with absolute ratio of 3/2 is best possible. We also prove that the Greedy algorithm is 3/2-competitive in both the absolute and asymptotic cases. For the problem without the assumption, i.e. calls may terminate at arbitrary time, we give the lower bounds for the competitive ratios: the absolute one is 5/3 and the asymptotic one is 14/9. We propose an optimal online algorithm with both competitive ratio of 5/3, which is better than the Greedy algorithm, with both competitive ratios 2.

On-Line Multiple-Strip Packing

Virtualization brings many benefits such as improving system utilization and reducing cost through server consolidation. However, it also introduces isolation problem when running multiple virtual machine workloads in one physical platform. Additionally, with the advent of multi-core technology, more and more cores are built into one die in todays data center that will share and compete for the resource like cache. Its worthy to study the isolation of server consolidation in modern multi-core platform. However, to our knowledge there are few work done on the isolation property especially the fault isolation property when one of the virtual machine workloads is attacked in server consolidation. In this paper, we study the isolation property from performance perspective and provide two optimization methods to improve the isolation property. We first define the isolation property and quantify the performance isolation in consolidation and propose a VM-level optimization method. Then we study the fault isolation by introducing a misbehavior virtual machine in server consolidation scenario and propose a core-level cache-aware optimization method to improve the fault isolation. Experimental results show that our two optimization methods can effectively improve the performance isolation and fault isolation with 29.39% and 19.52% respectively. Whats more, Oprofile/Xenoprof toolkits are used to find out the factors affecting isolation property from the hardware events level.

Efficient algorithms for finding a longest common increasing subsequence

Given a mobile telephone network, whose geographical coverage area is divided into cells, phone calls are serviced by assigning frequencies to them, so that no two calls emanating from the same or neighboring cells are assigned the same frequency. Assuming an online arrival of calls and the calls will not terminate, the problem is to minimize the span of frequencies used. By first considering χ-colorable networks, which is a generalization of (the 3-colorable) cellular networks, we present a (χ+1)/2-competitive online algorithm. This algorithm, when applied to cellular networks, is effectively a positive solution to the open problem posed in [8]: Does a 2-competitive online algorithm exist for frequency allocation in cellular networks? We further prove a lower bound which shows that our 2-competitive algorithm is optimal. We discover that an interesting phenomenon occurs for the online frequency allocation problem when the number of calls considered becomes large: previously-derived optimal (lower and upper) bounds on competitive ratios no longer hold true. For cellular networks, we show new asymptotic lower and upper bounds of 1.5 and 1.9126, respectively, which breaks through the optimal bound of 2 shown previously.

The Range Assignment Problem in Static Ad-Hoc Networks on Metric Spaces

We study the multiple-strip packing problem, in which the goal is to pack all the rectangles into m vertical strips of unit widths such that the maximum height among strips used is minimized. A number of on-line algorithms for this problem are proposed, in which the decision of delivering the rectangles to strips as well as packing the rectangles in strips must be done on-line. Both randomized and deterministic on-line algorithms are investigated, and all of them are guaranteed to have constant competitive ratios.