ARM on the Server

A picture of the Amazon Graviton ARM Chip

What a week!  Several of my predictions have come true but the most important of which, and the one which I think is going to have the largest impact is Arm on the Server.

Amazon this week announced the availability of A1 instance types, at a considerable cost saving.  The CPU’s are about 40% cheaper, though aren’t as performant right now, however we believe the actual energy saving will be close to 60%, leaving Amazon a 20% profit margin by being first to market.  Considering the R&D required by Amazon to create the custom chips (originally to be supplied by AMD), this is a fair return on investment and balance of cost saving passed on to the consumer.

For suitable workloads this represents quite a big cost saving.  While it’s true a CPU bound workload will be cheaper on an x86 chip, it’s worth noting that not all workloads are CPU bound and for them the wastage of having a CPU sat idle is not cost effective.

For those of us in the know, who’ve been preparing for this for a long time this represents a great opportunity.  We are probably one of very few experts at delivering ARM workloads on the server at scale.  This will no doubt leave the rest of the cloud providers and industry professionals scrabbling to catch up.

What I would sincerely recommend to companies wishing to deploy products to servers or migrate existing products to take advantage, do is to get in touch sooner rather than later so we can start expanding to meet the new demand this will create over the coming months.

The A1 instances are available in five sizes, all EBS-Optimized by default, at a lower cost:

Instance Name vCPUs
EBS Bandwidth Network Bandwidth On-Demand Price/Hour
US East (N. Virginia)
a1.medium 1 2 GiB Up to 3.5 Gbps Up to 10 Gbps $0.0255
a1.large 2 4 GiB Up to 3.5 Gbps Up to 10 Gbps $0.0510
a1.xlarge 4 8 GiB Up to 3.5 Gbps Up to 10 Gbps $0.1020
a1.2xlarge 8 16 GiB Up to 3.5 Gbps Up to 10 Gbps $0.2040
a1.4xlarge 16 32 GiB 3.5 Gbps Up to 10 Gbps $0.4080

We’re experts at scale-out computing, perfect for ARM workloads.  Get in touch if you’d like to discuss more how we can help with your individual deployments.

Leave a Reply

Your email address will not be published. Required fields are marked *