Cost of Running Classification

These are the primary cost components in running the Acante service:

• Compute: the Acante lambda Sampler service is currently provisioned at 1 vCPU. The short-lived lambda cost should be minimal.
• S3 Buckets: the data volume will be less than 1GB, this cost should be minimal
• SQS Queue: this cost should be minimal
• Network Bandwidth: the meta-data such as classification results are transferred via S3 Sync
• S3 Sync: the meta-data such as classification is < 1GB, this cost should be minimal
• AWS Macie: primary cost item. We have configured an upper limit on this
• One-time cost of Backward Scanning: Macie costs can be capped at a predefined level and sampling levels are adapted to fit within the cap. Macie is free up to 1GB of data scanned, and $1/GB after that. Average sampling rates below <0.5% are practical with Acante's patented Adaptive Classification Engine massively reducing costs while maintaining high accuracy.
• Ongoing cost of Forward Scanning: Over time, Acante samples and classifies only objects that have changed. This ongoing cost is typically at least 1 order of magnitude lower than the One-time cost above.