MMS • Steef-Jan Wiggers
Article originally posted on InfoQ. Visit InfoQ
Recently Microsoft announced the general availability (GA) of new storage-optimized Azure Virtual Machines (VMs). These VMs are the Lasv3 and Lsv3 series designed to run workloads requiring high throughput and IOPS, including big data applications, SQL and NoSQL databases, distributed file systems, and data analytics engines.
The Lasv3 and Lsv3 VM series provide faster processors, increased networking, and higher remote disk throughput compared to prior generation storage optimized VMs. More specifically:
Furthermore, high-throughput, low-latency, and directly mapped local Non-Volatile Memory Express (NVMe) storage are features of the new Lasv3 and Lsv3 VMs. In addition, the VMs are available in capacities ranging from 8 to 80 vCPUs – and there are 8 GiBs of memory per vCPU and one 1.92TB NVMe SSD device per 8 vCPUs, with the most extensive VM sizes offering up to 19.2TB (10×1.92TB).
Zach Christopherson, Staff Performance Engineer, Databricks, said in a Microsoft Tech Community blog post about the new VM series:
At Databricks, we’ve evaluated the Lsv3 and Lasv3 Azure VMs for the lakehouse using a comprehensive series of benchmarks that reflect common and critical patterns present in real-world analytics workloads. The v3 L-series’ local NVMe devices and faster CPU options enable Databricks Photon, the native vectorized query engine, to deliver leading price/performance for the most demanding data warehousing workloads.
Note that Microsoft’s competitor in the cloud space, AWS, also provides storage-optimized VMs with Amazon EC2 I3 Instances.
Currently, the Lsv3 VMs are available in more than 20 regions. The Lasv3 VMs are initially available in Central US, East US, South Central US, and West Europe regions, with more to follow. And lastly, the pricing details for the VMs are available on pricing pages for Windows and Linux.