iAdmin, a developer of web filtering technologies, announces the release of a new version of the iAdmin URL SDK, a solution that allows you to integrate a category database covering all websites worldwide on a second-level domain into third-party content filtering software.
The main difference in this version is a new operating mode that does not require the entire category database to be loaded into RAM. This version can be used directly on hardware devices with limited memory. Category data will be cached in memory from disk the first time a specific site is accessed.
The iAdmin URL SDK was originally designed for use in streams that directly analyze network traffic and network packets, where latency is critical. The SDK will run several times faster than any DBMS or external cloud service. Consequently, it will require fewer resources. All calculations, memory allocation, and disk I/O operations are optimized for categorization tasks exactly as required! The in-memory cache requires a maximum of 77 bytes per site. For example, with 3 million unique site visits per month at a large company, 231 MB of physical memory will be required.
Performance tests of the new version have been conducted, and you can run them yourself.
| Intel XEON E5-2420 1.9 GHz, DDR3, SSD RAID 1 System | Intel XEON E5-2420 1.9 GHz, DDR3, SSD RAID 1 System | |
| 100,000 websites | 1 thread | 4 threads |
| 1st pass with cache loading from disk | Time 107 sec Speed 1 ms\site | Time 38 sec Speed 0.38 ms\site |
| 2nd pass, memory search only | Time 1.46 sec Speed 14 µs\site | Time 0.359 sec Speed 3.6 µs\site |
The SDK will be able to operate in large enterprise-scale networks without significant delays.
For high-load services, the full database caching in memory mode has been retained; however, it will require 13 GB of storage.
