Amazon FinSpace pricing

Managed kdb Insights compute can be used for the individual compute nodes in a dedicated cluster or for the hosts in a scaling group. You pay an hourly rate (billed at one-minute resolution), based on the quantity and size of nodes in your dedicated cluster or hosts in your scaling group. Note that only kx.sg* types work with scaling groups.

Note: The price per hour does not include kdb Insights software license. See here for details on kdb licensing.

Data is stored in kdb Databases using change sets. You pay for the amount of data in your database, which is the total of the amount of data you have added to a database via changesets. This is calculated by adding up the GB provisioned for each hour and dividing by the total number of hours in the month, resulting in a “GB-months” value.

Shared storage volumes can be configured to provide POSIX file system storage for writing and storing Tickerplant (TP) logs, writing and storing realtime database (RDB) savedown files, temporary scratch space, and faster database access via Dataviews and high-performance disk cache. You pay for the amount of volume storage configured. This is calculated by adding up the GB provisioned per hour and dividing by the total number of hours in the month, resulting in a "GB-months" value.

When used for cache on a dedicated cluster in a Multi-AZ cluster configuration, a cache in each AZ is automatically created, so your total cost will be 3 times the amount of cache storage you specify for the cluster via the API/Console.

Real-time clusters will persist data from memory to disk so that it can be added to the kdb historical database for historical analytics. For dedicated clusters you can configure separate storage, called Savedown Storage, for each node. For Scaling Groups, you can configure a Storage Volume, which is shared by all nodes in the cluster, to use for savedown storage. 

The amount of savedown storage required will depend on the volume of data your real-time cluster is processing each day. You pay for the amount of storage configured for each node in the cluster. This is calculated by adding up the GB provisioned per hour per active node in the cluster and dividing by the total number of hours in the month, resulting in a "GB-months" value.

Consider a customer that wants to setup a Quant Research environment on AWS using Managed kdb Insights with 10 years of historical equity tick history and supporting 3 Quants.

To support this example, there will be a FinSpace kdb Database with 10 years of equity tick history totaling 100TB (100,000GB) and adding 50GB of new data at the end of each day (EOD). There will be one small FinSpace Scaling Group to run a kdb cluster to perform processing of the new EOD data received (kx.sg.2xlarge with 54GB of available memory). There will be one larger FinSpace Scaling Group (kx.sg.8xlarge with 216 GB available memory) that will provide each Quant with their own 64GB kdb cluster. To speed up analytics, there will be a high-performance cache, large enough for 2 years of the historical data (20TB = 20,000GB).

Monthly cost calculation (for a 30-day month)

Historical Database Storage: $2,640

• 100,000 GB x $0.0264 per GB-month = $2,640
• 50 GB new data x $0.0264 per GB-month = $1.32 of storage costs added each day to the historical database
  

EOD Data Ingest Processing:  $26.46

•  1 scaling group node x 1 hr per day x 21 trading days per month x $1.26 per hour (kx.sg.2xlarge) = $26.46

Quant Compute for running 3 kdb clusters: $1,666.72

• 1 scaling group node x 8 hrs per day x 21 trading days per month x $5.04 per hour (kx.sg.8xlarge) = $846.72
• 20,000GB database cache x $0.041 per GB-month (12MB/s shared volume) = $820.00

Total monthly cost: $4,333.18

•  $2,640 (Historical DB) + $26.46 (EOD ingest) + $1,666.72 (Quant Compute)

For more advanced examples, such as a real-time ticker plant with event processing, please reach out to your account team for assistance in estimating and optimizing the cost for FinSpace.