Thinking about using aws pgVecor as vectordb, but need please help calculating the cost based on word counts?
In other words, how much it will cost me for 5 million words in the AWS US-east-1 region?
Published Aug 5, 2024
To calculate the cost of using AWS pgVector as a vector database, you will need to consider the costs associated with storage and any additional services you might be using.
US-east-1
- Storage: The cost of storing the vector data.
- Compute: The cost of running the PostgreSQL instance.
- I/O Operations: The cost associated with read/write operations.
For simplicity here, let's focus on the storage and computing costs. These costs can vary depending on the specific instance type that you choose and the amount of storage needed for your specific use case.
Here’s a general approach to estimating the cost:
Step 1: Estimate Storage RequirementsAssume each word is represented as a 300-dimensional vector, which is a common size. Each float in a vector typically takes 4 bytes. For 5 million words:
Step 2: Storage CostAWS RDS PostgreSQL storage costs about $0.10 per GB-month in the US-east-1 region.
Step 3: Compute CostThe compute cost depends on the instance type. Let’s assume you choose a db.t3.medium instance, which costs around $0.0416 per hour.
Total Monthly Cost / Additional Costs:
• I/O Costs: Depending on your read/write patterns, there may be additional costs for I/O operations.
• Backup Storage: Costs for automatic backups and snapshots.
• Backup Storage: Costs for automatic backups and snapshots.
The estimated cost here is approximately $30.55 per month. Additional costs for I/O operations and backups may also apply based on your specific use case and usage patterns.