Macrometa Pricing Reference
Related Article - Macrometa Subscription Tiers
API Operations
Priced per million database operations. $1.00 /1M Ops
The Macrometa GDN is an API-driven platform. All read, write, and process operations are counted towards API Operations billing.
Storage
Priced per GB per month. $0.25 GB/Month
A global collection will be charged for data storage in each PoP (region) in the Fabric.
A local collection will only be charged for data storage in that single PoP (region).
Data at rest has been compressed which reduces the storage size.
Example: 4GB of data is inserted into a global collection in the GDN PaaS (9 PoPs). After replication, the data is 4GB per PoP, this equals $1 per PoP. The invoice would be $9 per month.
CPU - Query Runtime
Price per hour, $.25/Hour.
CPU usage is a measure of the runtime in milliseconds. All CPU runtime is aggregated into a single number.
This is important because queries can have a huge range of runtime depending on the data model and complexity of the query.
Example: A query to retrieve a single record from a Key-Value collection requires much less time than a complex search query across multiple document collections.
CPU - Event Processing
*Release scheduled for 0.18.0*
Price per hour, $.40/Hour.
CPU usage is a measure of the runtime in milliseconds. All CPU runtime is aggregated into a single number.
RAM
Price per GB per hour. $0.15 GB/Hour
RAM is the most complex billing metric on the platform. It is currently used for collection indexes.
When an index is created it uses RAM to hold a subset of data, making it much faster to access. This allows for the optimization of query patterns. However, as a resource, RAM is much more expensive than storage on the disk.
Indexes are persistent and dynamic, this is why we bill using the amount of data per hour. Storage volume is aggregated and recorded each hour.
Stream Data Transfer In/Out
Priced per GB. $0.05 /GB
For each GB into and out of a stream, we charge $0.05. It is important to note that because the data in and data out may not always be equal depending on which features of streams you use.
Example:
You are streaming 100 GB that filters the data down to 10GB for consumers. The input charge would be $5 the output charge will be $.50. It would work the other way as well, you could input 10GB for $.50 and “fan-out” that information to 10 consumers which would be 100GB at $5.
Network Transfer (Egress)
Priced per GB. $0.02 /GB
Data transferred from the GDN to an IP Address outside of the fabric's network.
We don’t charge for ingress or data replicated or transferred within the GDN.
GDN sections usage:
Collections - Storage, API
Collection indexes - RAM
Queries - CPU, API
Streams - Stream Data Transfer(Throughput* I/O), API
Stream Workers - CPU, RAM*
Search - RAM, API
Graphs - Storage and RAM, API
*CEP limits, from 18.0 release
FAQ
How can I estimate the cost of a query?
In the GDN console, you can view the Query Profile after a query is executed. This profile contains the query execution time in milliseconds. The query execution time is billed at $0.25 per hour. No additional time is added, The total cost of the query will be 1 API operation ($1/Million) and the CPU time to run the query.
Example:
1 API operation = $0.000001
100 ms Query execution time = $0.00000694
Total Cost = $0.00000794
Difference between egress transfer and stream data transfer out?
If someone runs a query as a 3rd party user to get data from the system, that is egress transfer.
If a Macrometa user uses streams to catch the data from a collection, it's a process of going from collection to stream workers and back to the collection, circulating data that is not leaving the GDN, only Stream I/O charges apply.
As we stated above, egress is for all the data that goes out of the GDN. Streams are internal and all the moving data is inside the GDN. However, if data goes into the stream and out of the system, both charges would apply.
An example of usage of both processes:
Using a stream worker to backup data to S3.
Comments
0 comments
Please sign in to leave a comment.