News

Microsoft Expands Azure Compute Offerings with Two New VM Types

• By Kurt Mackie
• December 12, 2017

Microsoft added two new virtual machine (VM) families to its lineup of Azure compute options this month: the lower-cost B-Series and the memory-intensive M-Series.

Both new VM types are now generally available. The M-Series is particularly notable for having "the largest CPU and memory sizes, with up to 4 TB of RAM," according to Microsoft's announcement.

The company this month also previewed the Lv2-Series of Azure VMs, which uses AMD's Epyc processors for "demanding workloads like MongoDB, Cassandra and Cloudera" that are storage and I/O intensive.

The Lv2-Series supports up to 64 virtual CPUs, 512 gibibytes (GiB) of memory and 8 x 1.9TB of local solid-state disk (SSD) storage. Microsoft right now is just offering "early access" to the Lv2-Series, which requires filling out a survey form at this page.

The B-Series (for "burstable") comes in six sizes and is commercially available now in 19 regions. This series uses local SSDs (up to 64GiB) for storage, with options to use Intel Broadwell or Haswell E5-2673 processor chips. The B-Series VMs are billed by Microsoft as its "lowest cost option for customers with flexible vCPU requirements." Microsoft also described the B-Series as being designed to support Web servers and small databases, or it can be used in dev-test environments with predominantly low CPU use. B-Series dev-testers can build up credits for CPU spikes of short durations, Microsoft's announcement suggested. CPU credits accrue when the VM uses less than its baseline capacity.

The M-Series (for "memory") is commercially available now in four sizes in the "West US 2, East US 2 and West Europe" regions, with expansion happening "in the coming months." The VMs in this series use Intel Xeon E5-2673 processors. Use options include "extremely large databases or other applications that benefit from high vCPU counts and large amounts of memory," such as running SQL Hekaton or SAP HANA in-memory workloads. The M-Series supports up to 128 vCPUs, 3,800GiB of storage and 4,096GiB of local SSD storage. The series also features service-level agreement guarantees.

Microsoft's VM nomenclature follows a pattern where the series letter has a specific meaning and there are various VM generations available, according to an online video explanation by Aidan Finn, a Microsoft Most Valuable Professional and expert in VMs and cloud computing. He noted that picking the right Azure VM depends on understanding how it will be used, which will determine the processor, RAM and IOPS needs.

The A-Series is at the low end of the Azure VMs and is suitable for file servers, according to Finn. It's an entry point for small-to-medium businesses. The A-v2 version of the product is actually cheaper than the standard A-Series offering, he noted.

The D-Series is good at "disk activity" using the Intel Xeon processor, while the E-Series is designed to provide "extra RAM." The F-Series is good for application or Web servers, he added. The H-Series is for "high performance," while the G-Series is for "goliath." The M-Series, for "massive," supports up to 128 cores and 2TB of RAM, Finn noted.

The NV-Series uses Nvidia's chips, and there are three options: NV (virtualization), NC (compute) and ND (deep computing). There also is an LS-Series that runs on the G-Series hardware but it's "rare," Finn said.