Resource allocation and scheduling strategies using utility and the knapsack problem on computational grids
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Computational grids are distributed systems composed of heterogeneous computing resources which are distributed geographically and administratively. These highly scalable systems are designed to meet the large computational demands of many users from scientific and business orientations. This dissertation address problems related to the allocation of the computing resources which compose a grid. First, the design of a pan-Canadian grid is presented. The design exploits the maturing stability of grid deployment toolkits, and introduces novel services for efficiently allocating the grids resources. The challenges faced by this grid deployment motivate further exploration in optimizing grid resource allocations. The primary contribution of this dissertation is one such technique for allocating grid resources. By applying a utility model to the grid allocation options, it is possible to quantify the relative merits of the various possible scheduling decisions. Indeed, a number of utility heuristics are derived to provide quality-of-service policies on the grid; these implement scheduling policies which favour efficiency and also allow users to intervene with urgent tasks. Using this model, the allocation problem is then formulated as a knapsack problem. Formulation in this manner allows for rapid solution times and results in nearly optimal allocations. The combined utility/knapsack approach to grid resource allocation is first presented in the allocation of single resource type, processors. By evaluating the approach with novel utility heuristics using both random and real workloads, it is shown to result in efficient schedules which have characteristics that match the intended policies. Additionally, two design and analysis techniques are performed to optimize the design of the utility/knapsack scheduler; these techniques play a significant role in practical adoption of the approach. Next, the utility/knapsack approach is extended to the allocation of multiple resource types. This extension generalizes the grid allocation solution a wider variety of resources, including processors, disk storage, and network bandwidth. The general technique, when combined with new heuristics for the varied resource types, is shown to result in improved performance against reference strategies. This dissertation concludes with a novel application of the utility/knapsack approach to fault-tolerant task scheduling. Computational grids typically feature many techniques for providing fault tolerance to the grid tasks, including retrying failed tasks or replicating running tasks. By applying the utility/knapsack approach, the relative merits of these varied techniques can be quantified, and the overall number of failures can be decreased subject to resource cost considerations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it