Instances›AWS EC2›i3›i3.16xlarge
Compute
vCPUs64
Memory488 GiB
Mem / vCPU7.63 GiB
ProcessorXeon E5-2686 v4
Clock speed2.3 GHz
Architecturex86_64
Storage
Type8 × 1,900 GB NVMe SSD
Bandwidth14,000 Mbps
Throughput1750 MB/s
IOPS65,000
Networking
Performance25 Gigabit
Max ENIs15
IPv6Yes
GPUNone
Family sizes
i3.large2 vCPU · 15.25 GiB
i3.xlarge4 vCPU · 30.5 GiB
i3.2xlarge8 vCPU · 61 GiB
i3.4xlarge16 vCPU · 122 GiB
i3.8xlarge32 vCPU · 244 GiB
i3.16xlarge64 vCPU · 488 GiB
i3.metal64 vCPU · 488 GiB
i3.16xlarge

i3.16xlarge is AWS's storage-optimized EC2 size at 64 vCPU and 488 GiB. In the name, "i" is I/O, meaning local storage, "3" is the generation and "16xlarge" is the size within the i3 family. It runs on Intel Xeon E5-2686 v4 (Broadwell) at 2.3 GHz, and AWS introduced it in November 2016. AWS has since shipped i4i, i7i and i8g; i4i.16xlarge is the direct replacement at this size, scoring 1,102,498 CoreMark/s against this instance's 853,447 — roughly 29% more throughput for the same vCPU count. The closest Spot shape is mh.vs1.2xlarge at 16 vCPU, so this instance maps to 4 of them — 64 vCPU and 512 GiB across the fleet, $4.99/hr on AWS in US East (N. Virginia), from $0.084/hr on Rackspace Spot. Spot delivers compute as managed Kubernetes, so a node pool of that size is the native shape of the answer rather than a workaround — provided the workload scales horizontally.

Match qualityScale-out match

i3.16xlarge gives you 64 vCPU and 488 GiB in a single machine. The largest comparable Spot node, mh.vs1.2xlarge, is 16 vCPU and 128 GiB, so the equivalent here is 4 nodes rather than one.

Spot runs as managed Kubernetes, so a 4-node pool is the normal way to run things rather than a workaround. If your workload already sits behind a load balancer, a queue or a scheduler, it spreads across them without changes. If it needs all 64 cores and 488 GiB in one process, it does not.

No local diski3.16xlarge comes with 8 × 1,900 GB NVMe SSD attached to the machine. Spot nodes use network storage only, so that disk doesn't come with it. If you were using it as fast scratch space, you'd move to a persistent volume or object storage, which are slower to read.
Pricing by region
1 hr
RegionAWS on-demandAWS Spot1-yr Reserved3-yr ReservedRackspace SpotYou save
US East, N. Virginia$4.992$1.710$3.427$2.419$0.084
$4.908 savings
US Central, Chicagono AWS region$0.084
US Central, Dallasno AWS region$0.524
Hong Kong$6.592$4.516$3.198$0.084
$6.508 savings
Sydney, Australia$5.984$1.058$4.099$2.903$0.084
$5.900 savings
London, UK$5.792$1.467$3.965$2.834$0.524
$5.268 savings
US West, San Jose$5.504$1.500$3.763$2.696$0.008
$5.496 savings
US Central, Dallas 2no AWS region$0.968
When to use it
Good for
  • ●Distributed file systems and search indexes
  • ●Transactional databases with heavy random I/O
  • ●Streaming ingest that spools to local disk
  • ●Workloads that fit inside 488 GiB of RAM
  • ●Jobs that want 8 × 1,900 GB of local NVMe scratch space
  • ●Network-heavy services — AWS rates this size at 25 Gigabit
  • ●Anything that shards cleanly across 4 nodes, which is how it maps onto Rackspace Spot
Avoid if
  • ●Starting a new deployment. i4i.16xlarge is the newer generation at this size.
  • ●The working set needs more than 488 GiB of RAM
  • ●More than 7.63 GiB of RAM per vCPU is needed — that ratio is fixed for the whole family
  • ●The work is single-threaded. Most of these 64 vCPU would sit idle.
  • ●GPU or FPGA acceleration is needed
  • ●Workloads that never touch local disk. The instance-store premium is wasted.
i3.16xlarge runs $3,644/mo on AWS on-demand. The 4-node equivalent on Rackspace Spot: $61/mo.
  • Spot sells spare capacity at market prices. It is the cheapest compute on the internet.
  • Prices are set by supply and demand. You pay what the market decides.
  • Compute is delivered as fully managed Kubernetes. All infrastructure is taken care of.
  • Egress costs $0.01/GB after the free allowance. AWS charges $0.09/GB.
Build your Cloudspace
Frequently asked questions