Amazon Web Services (AWS) has introduced a new type of general purpose instance for its Elastic Compute Cloud (EC2).

The M4 instances are built around custom 2.4 GHz Intel Xeon E5-2676 v3 Haswell processors, and promise higher bandwidth and lower latencies when compared to the rest of the general purpose range.

They are aimed at a wide variety of use cases including relational and in-memory databases, gaming servers, caching fleets, batch processing, and business applications like SAP and Microsoft SharePoint.

“With these capabilities, M4 is one of our most powerful instance types and a terrific choice for workloads requiring a balance of compute, memory, and network resources,” said Matt Garman, vice president of Amazon EC2 at AWS.

The new instances are available immediately across eight Availability Zones in the US, Europe and Asia.

You can have any size, as long as it’s large

M4 comes in five different sizes ranging from two virtual CPUs with 8GB of memory (m4.large) to 40 vCPUs and 160GB of memory (m4.10xlarge).

The new instances enjoy dedicated bandwidth to Amazon Elastic Block Store (Amazon EBS) – a persistent, high availability storage platform for EC2.

More importantly, these are the first General Purpose instances to get support for ‘Advanced Networking’ feature set – previously available as part of specialized EC2 instances like C3, C4, D2, I2 and R3. Advanced Networking enables an instance to virtualize its Network Interface Card to send and receive traffic without the involvement of the Xen hypervisor.

AWS claims this approach helps deliver up to four times the packet rate of older instances. And for resources in the same Placement Group - a logical collection of instances within a single Availability Zone - Enhanced Networking reduces average latencies between instances by 50 percent or more.

Marketing software developer HubSpot was among the first to test the new virtualized infrastructure. “M4 instances offer an optimal balance of compute and memory for our cluster, and the m4.10xlarge, with 40 vCPUs and 160 GiB of memory, will allow us to significantly reduce our cluster size while driving down costs through better hardware utilization,” said Whitney Sorenson, vice president of Platform Infrastructure at HubSpot.