Moohyun Song

Taeyoon Kim, Kyumin Kim, Kyunghwan Kim et al.

Cloud vendors offer discounted spot instances to maximize surplus resource utilization, but these instances are subject to the risk of sudden interruption. Traditional pricing datasets have been employed to predict this risk, yet recent policy changes by cloud vendors have diminished their effectiveness. To promote spot instance usage, public cloud vendors provide instant availability datasets to help users mitigate interruption risks. While existing research utilizing this data has proposed methods to reduce interruptions, these studies have primarily focused on single-node instances, overlooking the stability of multi-node environments widely adopted for modern cloud workloads. This paper proposes SpotVista, a system that recommends a resource pool of reliable and cost-efficient multi-node spot instances by leveraging various publicly available datasets. To achieve this, SpotVista collects a large-scale multi-node availability dataset while overcoming significant query limitations. Through a thorough analysis of multi-node spot instance availability behavior, SpotVista establishes a methodology for recommending cost-efficient and reliable multi-node configurations. To evaluate how effectively the proposed methodology reflects multi-node availability and cost efficiency, extensive real-world interruption experiments were conducted. The results demonstrate that SpotVista outperforms the state-of-the-art work, SpotVerse, achieving 81.28% greater availability and 2.84\% more cost savings in a multi-region setup. When compared to a publicly available service, AWS SpotFleet, SpotVista provides 21.6\% higher stability and 26.3% greater cost savings.

Kyumin Kim, Moohyun Song, Taeyoon Kim et al.

Spot instances offer significant cost savings of up to 90% over on-demand prices, making them an attractive resource for large-scale computing workloads. However, understanding their availability dynamics is essential for building systems that tolerate interruptions, and observing this availability directly requires keeping instances running, which incurs costs that scale with the number of monitored instance types and their per-instance price. We propose Spot-and-Scoot (SnS), a cost-efficient method that collects spot instance availability signals by leveraging the cloud provider's provisioning lifecycle. Since the outcome of a spot request is determined before the instance enters the running state, SnS submits requests and cancels them upon provisioning acceptance, collecting binary availability signals at near-zero instance cost. Submitting multiple concurrent requests per measurement point further yields a quantitative estimate of available capacity. We validate SnS through simultaneous collection of probing signals and actual running instance traces across 68 instance types and 15 regions on both AWS and Azure, totaling 336,033 spot requests. Analysis of 2,635 real-world interruption events reveals that co-interruptions within the same instance type and availability zone occur within three minutes in over 92% of cases, motivating a binary availability formulation. Based on this formulation, we derive three complementary features from SnS signals and demonstrate that their combination achieves an F1-macro score of up to 0.90 for current availability modeling and maintains 0.85 at a 60-minute prediction horizon. A trace-driven simulation using TPC-DS workloads further demonstrates the potential of SnS-based prediction to reduce lost computation compared to an unguided baseline.

Seungwoo Jeong, Moohyun Song, Juhyun Park et al.

As large language model (LLM) services become widely adopted, the cost of GPU resources for serving these models in cloud environments has emerged as a critical concern. Spot instances offer up to 90% cost savings over on-demand instances, but their frequent interruptions and limited availability pose significant challenges for continuous LLM serving. GPU spot instances, in particular, exhibit lower and more volatile availability than CPU-based instances, making homogeneous clusters that depend on a single GPU type vulnerable to correlated failures. Heterogeneous clusters spanning multiple GPU types can address this by leveraging complementary availability patterns across diverse spot pools, yet existing LLM serving systems are designed for homogeneous environments and suffer from load imbalance when deployed on heterogeneous GPUs. This paper presents ShuntServe, a cost-efficient LLM serving system for heterogeneous spot GPU clusters. ShuntServe employs a roofline model-based analytical serving performance estimator and a dynamic programming-based model placement optimizer that jointly determines node configuration, parallelization strategy, and layer assignment to maximize throughput across heterogeneous GPUs. To enhance fault tolerance when using spot instances, ShuntServe combines output-preserving request migration with concurrent initialization via a shared tensor store, minimizing migration downtime by overlapping replacement node preparation with ongoing serving. Evaluation on Llama-3.1-70B and Qwen3-32B with a heterogeneous AWS cluster of L4, A10G, and L40S GPUs shows that ShuntServe achieves 1.42x and 1.35x higher throughput than state-of-the-art baselines and attains 31.9% and 31.2% cost efficiency improvements over on-demand instances for offline and online serving, respectively.