Intel recently showed off a range of advanced-stage research projects to the press and potential partners at its fourth annual Research at Intel Day held at its facility in Santa Clara. The projects spanned mobile technology, enterprise computing, large-scale computing platforms and "people-centred" computing and are being developed under Intel's current mantra of "driving energy efficiency and performance."
In his keynote session that opened that day, Justin Rattner, Intel CTO, said: "It takes a good four years to develop a new generation of microprocessor, and another 3 or 4 years preceding that for research and getting the ideas. What you will see today is work that has been going on for the last few years." Intel did not announce which of the projects would make their way into the market.
Rattner said one of Intel's goal is to achieve 10 times improvement in the energy efficiency and performance of its processors over the next three to four years. In communications, he said the major theme for Intel researchers is WiMAX and ultrawideband, adding that the two technologies will be "fully deployed in the platform over the coming year." In enterprise computing, Rattner said Intel is "going after the maintenance portion of the pie," with research focused on virtualization, data center performance and security.
Among the enterprise computing research being demonstrated Wednesday, were:
The adaptive firewall
Intel's traffic-adaptive filtering technology has been in development for two years. It sits on any node on the network and learns about traffic patterns to introduce shortcuts to frequently travelled paths. In their demonstration, Intel researchers showed a video streaming application going from a server to a client via a router with a firewall. The researchers launched a denial-of-service attack against the router but the video traffic was unaffected because the filtering technology had placed shortcuts in the frequently travelled paths between the server and the client, and the attacker to firewall, which reduced the number of memory accesses in the classification process and increased the throughput of the firewall. The researchers said they plan to make available the technology as open source by year-end.
Trusted platforms with virtualization
This research puts Trusted Platform Modules (TPM) into the virtualized computing environment. TPMs, based on specification developed by The Trusted Computing Group, are microcontrollers used to store and authenticate passwords, digital certificates and encryption keys. In its research, Intel puts software-based Virtual TPMs (VTPM) in front of each virtual machine client to attest their status to the authentication server, which decides to allow or deny the virtual machine's access to the whatever server it wants to connect to based on their status as reported by the VTPM. The technology has been in development for almost two years.
Dynamic thermal management of the data centre
Developed in conjunction with Arizona State University, this research enables job scheduler software to take into account the temperature of servers or server blades before deciding which data centre component should do the job. The result should be an online thermal control framework that monitors and manages data centre thermal performance from a holistic viewpoint. The researchers say the challenge for the project is to make the system reactive so that it knows when servers are starting to fail because of heat issues. They say it could be another two years before this project could be presented to Intel as a potential product.
Corroboration catches stealthy worms
Slow worms are hard to catch, say Intel researchers, particularly if they try to hide in background traffic. In distributed detection inference, a node would raise an alarm if something odd is happening and reports this to another node. However, the node may be reporting a false positive. But if a few other nodes do the same thing -- raise alarms and report to another node -- the message would eventually reach a trusted node, which would decide on what action to take. Intel researchers describe this as nodes "gossiping" for the good of the network and said that most of the time, the network is able to tell that a worm is attacking when there is a low-level of infection. Researchers say the challenge is to put this distributed detection inference technology onto the node's hardware rather than software, which would put it at risk of being taken over by the worm.
Intel researchers say the all-wireless mesh moves all network capabilities to the edge, enabling all nodes on the network to communicate with and through each other using wireless technologies and mesh networking. Intel said the technology could promote the formation of ad-hoc groups that could share different information between each other wirelessly. The research uses distributed virtualization based on PlanetLab, an experimental network built by academic and commercial researchers across the world that rides over the public Internet.