V2 BETA – Initial Performance MetricsD
ata Portcullis already performs with incredible
speed and efficiency – but the version currently in Beta Testing at our labs will simply blow you away.
We noticed that some Websites / Webservers may not need to operate Data Portcullis in real time to decide whether to reject an incoming connection – a connection may be rejected after a minute or two.
For example – if an attacker is attempting to book a flight, a hotel room or a concert ticket on your Server, this is an “end to end process” that will take a few minutes to conduct so the attacker can be rejected at any point during the end to end process, rather than upon their first connection to your Landing Page.
Data Portcullis Beta Version 2.0 introduces an optional ‘Latency Flag’ for those Websites / Webservers who can wait a few minutes to reject their attacker. Initial testing indicates that by introducing a slight latency of between 1 and 3 minutes on the final accept/reject decision increases connection throughput handling rates by tenfold.
2000 Connections Per Second!
The below diagram is a server monitoring snapshot of a ten minute run, showing that the average number of requests being handled is 2000 per second
, or 125,000 per minute.
This is a ‘snapshot’ of a real Data Portcullis Server and Database being stress tested at our labs using the ‘latency’ option. This ten minute long snapshot shows the average throughput rate handling around 2000 connections per second.
Average throughput – 125,000 requests per minute (2000 requests per second)
Here are a few of the system stress metrics taken during the above test. The most interesting metric can be seen bottom left, where the production server processors barely reach 75% of their available workload capacity – even while processing 2000 transactions per second.