# Azure previews Lasv5 and Laosv5 local-storage VMs

> The AMD EPYC Turin-based VM families increase local NVMe capacity, network bandwidth, size range, and processor performance for storage-heavy workloads.

Canonical URL: https://www.devobs.io/news/news-azure-lasv5-laosv5-amd-turin-preview/
By: Samira Haddad
Published: 2026-09-06T11:58:54.638Z
Updated: 2026-09-06T11:58:54.638Z
Event date: 2026-06-02
Section: Infrastructure

Azure previewed Lasv5 and Laosv5 local-storage-optimized virtual machines powered by fifth-generation AMD EPYC “Turin” processors. The [Compute announcement](https://techcommunity.microsoft.com/blog/azurecompute/announcing-preview-of-new-azure-lasv5-and-laosv5-vms-based-on-the-amd-epyc%E2%84%A2-%E2%80%98tur/4522407) presents the families for workloads that need high local disk throughput, with configuration choices suited to data processing, caching, and storage engines.

## Local storage is a topology decision

New CPU and NVMe generations can improve a node's ceiling, but local disks remain tied to the VM host. They fit replicated databases, shuffle space, indexes, or caches whose contents can be rebuilt. They are a poor sole home for durable data unless the application replicates and verifies that data elsewhere.

Lasv5 and Laosv5 differ in resource balance and storage arrangement, so family names alone are insufficient for sizing. Preview availability also varies by region and size. Microsoft’s performance figures are starting points, not workload guarantees.

## Test durability before speed

Choose a representative dataset and benchmark ingestion, compaction, query latency, recovery, and network replication together. Record CPU frequency behavior, memory pressure, local disk queues, and remote storage traffic. Compare against the current production family with identical software and redundancy.

Then force node loss or redeployment. Measure rebuild time and the load it places on peers and remote storage. Confirm that the service maintains quorum and capacity while a large local dataset returns. Automate disk discovery and formatting because device layout may differ from older images.

Before adoption, check preview terms, regional capacity, quota, guest support, and maintenance behavior. A compelling steady-state result should advance only when the workload also meets its recovery objective after losing all data on one VM's local disks.

## Source references

- <https://techcommunity.microsoft.com/blog/azurecompute/announcing-preview-of-new-azure-lasv5-and-laosv5-vms-based-on-the-amd-epyc%E2%84%A2-%E2%80%98tur/4522407>