Dynamic, Hierarchical Resource Allocation with Throttling for vSAN
This project focuses on performance isolation in VMware vSAN, using a dynamic linear programming model to optimize traffic allocations across virtual and physical disks while subject to both SLA and device capacity constraints. This approach takes advantage of the EXSi hypervisor's capabilities to throttle individual VMs, avoid heavy global data movement, and effectively isolate burst traffic.
Abstract
Performance isolation is critical in multi-tenant systems like VMware vSAN. We propose a hierarchical IOPS allocation design: tenants receive an overall IOPS quota, which is subdivided among its virtual disks, each striped over a fixed set of physical devices. We present a linear programming formulation to periodically recompute the rate allocation for each virtual disk, optimized for total system throughput, while subject to the estimated virtual disk demands, tenant-level SLAs, and the capacity limits of the underlying physical devices. Taking advantage of the vSphere hypervisor’s throttling mechanism, our simulation shows this approach avoids heavy global data movement, effectively isolates burst traffic, reduces congestion, and only incurs small runtime overhead by distributing the computation.