AWS Documentation » AWS Snowball » Developer Guide » Using an AWS Snowball Edge » Using Amazon EC2 Compute Instances » Limits for Compute Instances on a Snowball Edge

This guide is for the Snowball Edge. If you are looking for documentation for the Snowball, see the AWS Snowball User Guide.

Limits for Compute Instances on a Snowball Edge

The following AWS Regions are supported for you to add compute instances to your devices:

• US East (Ohio)

• US East (N. Virginia)

• US West (N. California)

• US West (Oregon)

• Asia Pacific (Singapore)

• Asia Pacific (Sydney)

• Asia Pacific (Tokyo)

• EU (Frankfurt)

• EU (Ireland)

• EU (London)

• EU (Paris)

• South America (São Paulo)

Storage Limits

The storage available for compute resources is a separate resource from the dedicated Amazon S3 storage on a Snowball Edge device. The limits for storage are as follows:

• Storage limits for the Snowball Edge Storage Optimized option – The total available storage for Amazon S3 is between 60 TB and 80 TB depending on whether you're using compute instances on the device. If you are using compute instances, then total available dedicated storage for sbe1 compute instances for the Snowball Edge Storage Optimized option is 1,000 GB.

• Storage limits for the Snowball Edge Compute Optimized and with GPU options – The total available dedicated storage for sbe-c and sbe-g instances is 7.68 TB. The total available storage remaining is 42 TB.

The following tables outline the available compute resources for Snowball Edge devices.

Feature	Limitation
Number of AMIs on a single Snowball Edge Storage Optimized option	10
Number of AMIs on a single Snowball Edge Compute Optimized option	20
Number of AMIs on a single Snowball Edge Compute Optimized with GPU option	20
Number of volumes per instance	1
Concurrently running (or stopped) instances	Varies depending on available resources

Instance type	vCPU cores	Memory (GiB)	GPUs	Supported device option
sbe1.small	1	1	0	storage optimized
sbe1.medium	1	2	0	storage optimized
sbe1.large	2	4	0	storage optimized
sbe1.xlarge	4	8	0	storage optimized
sbe1.2xlarge	8	16	0	storage optimized
sbe1.4xlarge	16	32	0	storage optimized
sbe1.6xlarge	24	32	0	storage optimized
sbe-c.small	1	2	0	compute optimized with GPU
sbe-c.medium	1	4	0	compute optimized with GPU
sbe-c.large	2	8	0	compute optimized with GPU
sbe-c.xlarge	4	16	0	compute optimized with GPU
sbe-c.2xlarge	8	32	0	compute optimized with GPU
sbe-c.4xlarge	16	64	0	compute optimized with GPU
sbe-c.8xlarge	32	128	0	compute optimized with GPU
sbe-c.12xlarge	48	192	0	compute optimized with GPU
sbe-g.small	1	2	1	with GPU
sbe-g.medium	1	4	1	with GPU
sbe-g.large	2	8	1	with GPU
sbe-g.xlarge	4	16	1	with GPU
sbe-g.2xlarge	8	32	1	with GPU
sbe-g.4xlarge	16	64	1	with GPU
sbe-g.8xlarge	32	128	1	with GPU
sbe-g.12xlarge	48	192	1	with GPU

Shared Compute Resource Limitations

All services on a Snowball Edge device use some of the finite resources on the device. A Snowball Edge device with its available compute resources maximized can't launch new compute resources. For example, if you try to start the file interface while also running a sbe1.4xlarge compute instance on a storage optimized device, the file interface service doesn't start. The following outlines the available resources on the different device options as well as resource requirements for each service.

• If no compute services are ACTIVE:

  • On a storage optimized option, you have 24 vCPUs and 32 GiB of memory for your compute instances.

  • On a compute optimized option, you have 52 vCPUs and 208 GiB of memory for your compute instances. This is also true for the with GPU option.

• While AWS IoT Greengrass and AWS Lambda powered by AWS Greengrass are ACTIVE:

  • On a storage optimized option, these services use 4 vCPU cores and 8 GiB of memory.

  • On a compute optimized option, these services use 1 vCPU cores and 1 GiB of memory. This is also true for the with GPU option.

  • While the file interface is ACTIVE it uses 8 vCPU cores and 16 GiB of memory on a Snowball Edge device.

You can determine whether or not a service is ACTIVE on a Snowball Edge by using the command snowballEdge describe-service on the Snowball client. For more information, see Getting Service Status.