The Web filtering engine is designed to store website categories locally.
Artica provides a database of over 70 million websites categorized into more than 250 topics.
Web filtering operates by querying these databases, which are downloaded and stored locally on your server’s disk.
Depending on the categories you enable in your Web filtering rules, the service will decompress the databases and load the corresponding website datasets into memory.
The more categories you load, the higher the memory footprint will be.
There is a significant advantage: using local databases allows the system to determine a website’s category extremely quickly.
However, this performance comes at a cost — increased memory consumption.
We estimate that approximately 250 million websites are currently active worldwide.
This means the databases are continuously growing, increasing day after day.
¶ How reduce memory usage ?
To reduce memory usage, you need to decrease the number of categories loaded in your rules.
- Proxy ACLs, on the other hand, rely on DNS queries to the Artica cloud and offer the same capabilities as Web filtering by querying an external server over the Internet.
For example, if you consider that the malware category (one of the largest categories) is included in all your rules, you can remove it from Internet filtering and instead handle it within the proxy ACLs.
- Another approach, if you are using multiple proxy servers, is to deploy a centralized filtering server.
In this setup, your proxies will no longer run the filtering service locally but will query a dedicated filtering server instead.
Consolidating filtering onto a single server will effectively reduce the memory footprint across all proxies in your infrastructure.