API Gateway 10.11 | Administering API Gateway | Performance Tuning and Scaling | Scaling
 
Scaling
 
Scaling up API Gateway
Scaling up API Data Store
Scaling down API Gateway
Scaling down API Data Store
As a critical step in your API Gateway deployment, you perform the capacity planning for API Gateway and its components that can live up to the estimated transactions (TPS) demands and data volume storage needs in compliance with your data and analytics retention SLAs. Though it is recommended to have the right sizing in place, it is important to consider scaling up or scaling down API Gateway to address the spikes that are not factored in the initial capacity planning.
You can do horizontal scale up or scale down of API Gateway to meet the spikes in the transactions or data.
You can scale up:
*API Gateway. When the number of transactions exceeds the estimated capacity. For example, if the existing deployment handles 200 transactions per second, you have to scale up API Gateway if the number exceeds this limit.
*API Data Store (Elasticsearch). When the volume of data to be stored exceeds the estimated size. For example, if the API Data Store capacity is 500 GB, and if the data to be stored exceeds the estimated size, you have to scale up the API Data Store.
Note:
You can minimize the need for the scaling of API Data Store with the right sizing (capacity planning), monitoring, and Data housekeeping procedures that are in compliance with your data and analytics retention SLAs.
Scaling requirements analysis
You can monitor the key metrics of system and application to decide whether they must be scaled up or not.
*For information on monitoring API Gateway, see Monitoring API Gateway .
*For information on monitoring API Data Store, see Monitoring API Data Store or Infrastructure Metrics.
Software AG recommends that you
*Set up Data housekeeping procedures.
*Separate the lifecyle of API Gateway and Data store in a clustered environment.
*Scale up only the required component based on the need (API Gateway for transactions and Elasticsearch for Data volume).