How does ParkerDB.com help? #
The parker-preview docker image can provide the point query service with ultra-low latency. However, it is not production-ready with the following requirements:
- Auto Refresh: The data could be updated regularly. In this case, you need to refresh the data in the instances periodically, while still keeping the service available with ultra-low latency.
- Data Partitioning: The data could be too big to fit into one machine. In this case, you need to partition the data to multiple instances.
- Linear Scalability: If the data size grows, you need to add more instances and balance the data to handle the query load.
- High Availability: If a few servers are down, the service should still be available.
To solve these problems, you need Parker Admin hosted on ParkerDB.com to manage the Parker instances. The parker instances can be hosted on your own cloud provider or on-premises cloud.
What is ParkerDB architecture for BYOC? #
BYOC(Bring Your Own Cloud) is a great way to have full control over your data and the infrastructure.
architecture-beta group vpc(cloud)[VPC] service gateway(internet)[Load Balancer] in vpc group public_api(cloud)[Parker Cloud] service admin(server)[Parker Admin] in public_api group parker_instances(cloud)[Parker Instances] in vpc junction jLB in parker_instances junction jLBL in parker_instances junction jLBR in parker_instances service pi1(server)[Parker] in parker_instances service pi2(server)[Parker] in parker_instances service pi3(server)[Parker] in parker_instances gateway:B -- T:jLB jLB:B -- T:pi2 jLB:L -- R:jLBL jLB:R -- L:jLBR jLBL:B -- T:pi3 jLBR:B -- T:pi1 admin{group}:L --> R:pi1{group}
- Your VPC contains Parker instances.
- You can use a load balancer to randomly route the queries to the Parker instances.
- Parker instances communicate with each other to locate data.
- Parker Admin in the cloud helps to organize the data.
Parker Admin #
Parker Admin has the following features:
- Trigger the data refresh in Parker instances.
- Partition the data to multiple Parker instances.
- Replicate the data to ensure high availability.
- Balance the data to handle the query load.
Parker instances #
- Each Parker instance contains one portion of the data.
- Parker instances communicate with each other to locate the data.
- Parker instances can be added or removed to handle the query load.