Arbor Networks has launched a plan under which it hopes data centres and web hosts will be able to assist ISPs in thwarting flood-based distributed denial of service (DDoS) attacks.
The vendor today launched a "standalone" version of its threat management system (TMS) that it said would cost half that of the existing version (which was "tightly integrated" with the vendor's network visibility tool Peakflow).
The appliance "gives data centre operators and web hosts all the same features as the ISP version, but at a more cost-effective price," said Arbor's Australian managing director Nick Race.
Race said that data centre operators did not require the "network-wide visibility" afforded by Peakflow, which monitored switches and routers on ISP networks.
But "multi-tenant environments" like data centres and web hosts "are prime targets for DDoS attacks because of their high profile and the potential to cause collateral damage across multiple customers," Arbor said.
Last month, hundreds of NetRegistry customers became collateral damage after an attack on an anti-piracy website caused degraded performance on a load balancer that served a block of websites.
Arbor believed there was a market for data centre operators and web hosts to provide protection to customers against application layer DDoS attacks.
Race also said that getting Peakflow TMS tools into data centre architectures would help mitigate more common "volumetric" attacks which bombarded communication links.
"The nice thing about Peakflow being deployed so widely in ISPs is the ability for the data centre infrastructure to signal up to ISPs to assist in [battling] volumetric attacks," Race said.
"It's about self-defence [for the data centre] and being able to work in conjunction with telcos [and ISPs]."
Race said Arbor was working on getting TMS boxes into data centres locally, but could not reveal who was trialling the boxes.
Arbor also announced today the availability of version 5.5 of its Peakflow SP product, which included new infected host detection capabilities for recognising bot networks.