You can deploy NetScaler Insight Center either as a single server or as a scale-out deployment with multiple collectors and database nodes.
This document enables you to determine your requirements for a new deployment as well as plan the growth of an existing NetScaler Insight Center deployment.
The following sections provide an overview of the two deployment topologies:
In a NetScaler Insight Center single server deployment, the database is integrated with the server and provides limited storage space. Also, a single server processes all the traffic. The following diagram shows how a single NetScaler Insight Center server communicates with the NetScaler device.
This deployment is recommended when you do not have much data to store or when the number of HTTP or ICA requests per second is less. For example, you might have one NetScaler Gateway and 500 users and you want to store data for only 31 days. Or, the number of HTTP requests per second is less than 5000 and you want to view only real-time data. In such cases, a single server deployment is recommended.
If your requirements change at any point of time with an increase in the number of users, data, or requests per second, you can upgrade to the scale-out deployment.
In a NetScaler Insight Center scale-out deployment, you can add agents to increase the processing power and you can add multiple database nodes to increase the storage space. The NetScaler Insight Center server helps you decide the number of resources you might need in this deployment by implicitly calculating the number of recommended resources on the basis of the number of HTTP requests per second, number of ICA sessions, or number of active WAN connections in your environment.
You are recommended to use this deployment when you want to store a high volume of data, for example, per-minute data for one year. A single server has limited storage space and cannot store data for one year. In this case, you are recommended to provision multiple database nodes.
You might also consider deploying this when there is a high amount of traffic passing through your network devices, for example, 20,000 HTTP requests per second, resulting in a high quantity of data required to be decoded. A single server might not be able to process such high volume of data leading to slowness of the server and delay in real-time data reporting.
The following examples explain scenarios where you might have to provision a scale-out deployment.
- If the number of HTTP requests per second is 5000 but you want to store the data for one year, you might not need an agent, but it is recommended that you deploy multiple database nodes store the data.
- If the number of HTTP requests per second is 10,000, it is recommended that an agent be installed to process the data. Additionally, multiple databases are required to store the high volume of data.
- For ICA traffic coming from 500 users, you might not need an agent. But if you want to store the data for one year, you should install multiple database nodes.
In a NetScaler Insight Center scale-out deployment, in addition to the server, three more components – agents, connectors, and database nodes – are added to meet your performance and scalability requirements.
The following figure shows how a NetScaler Insight Center scale-out deployment works.
As shown in the above figure, the server and the agents receive the data records from the monitored devices, process these records, and send the data to the connectors. Then, the connector uniformly distributes this data across the database nodes.
When you install an agent and register it with a server, the agent is implicitly registered as a collector on all the monitored NetScaler devices. You can install multiple agents to form a distributed deployment for handling high amount of appflow data.
Connectors and database nodes work together. Connectors efficiently load balance the insertion and retrieval of data to and from the database nodes.
You can install multiple connectors to distribute the data in parallel. Citrix recommends that you install a minimum of two connectors for one or more database nodes.
You must assign a uniform size for each database node in a database cluster. You can install multiple database nodes to store a large amount of data.