Occasionally, the network connection between the Data Aggregator and the Data Collector hosts is lost. In this case, Data Collector continues polling and caches polled data in memory, up to a configurable limit. When the Data Aggregator host becomes available, the cached polled data is sent to Data Aggregator.
Polled data is processed in a "first in, first out" order. That is, the oldest cached polled data is sent to Data Aggregator first. If the cache memory limit is reached, any new polled data is lost until the Data Aggregator host becomes available and the Data Aggregator host has processed 9 percent of the cached data.
Important! Memory usage increases dramatically on the Data Collector system when Data Aggregator is unavailable.
Your memory storage requirements vary and depend on the following factors:
The default value for the cache memory limit is one half of the maximum Data Collector process memory. You configured the maximum memory usage when you installed Data Collector or after the installation.
Data Collector requires a dedicated amount of memory to function properly. In a small-scale environment, with Data Collector polling 50,000 devices and components at a five-minute poll rate, 2 GB of memory is required for basic operation. In a large-scale environment, with Data Collector polling 500,000 devices and components at a five-minute poll rate, 24 GB of memory is required for basic operation. The remaining memory can be used for caching polled data.
|
Copyright © 2015 CA Technologies.
All rights reserved.
|
|