Cybersecurity's best bet: An Internet that learns to defend itself

During the first 40 years of the Internet, security has been about patching mistakes that have been made in its architecture and software and defending these vulnerabilities against attacks from the outside. The future will be about creating an environment that is secure by design and has the ability to anticipate rather than merely respond to threats, government and industry experts say.

This will not be a single secure infrastructure, but an ecosystem of technologies designed to better handle the basic chores of systems development; identity management and access control; and monitoring, analysis and response.

The Internet was not designed to be secure. When originally launched by the Defense Department’s Advanced Research Projects Agency, it was not intended for public use. As it grew, development has always been more about what could be done with it rather than what should or should not be done. Today the Internet has become integral to the economy and security, while cybersecurity has become a front-page issue and observers see the need for fundamental shifts in the way we secure it.

“It is clear that old, reactive, perimeter-based models of security are inadequate.”
-- Art Coviello, RSA.

“This does not mean perfect security, but instead a model that evolves and learns from change, whether process, technology or threat related,” said Art Coviello, executive chairman of RSA.

“We have a tendency to talk about the threat, and there’s not a lot we can do about that,” said Ron Ross of the National Institute of Standards Technology, who heads the implementation program for the Federal Information Security Management Act.

Good cybersecurity hygiene is important, Ross said. That means covering the basics of knowing your systems, understanding and managing their configurations, and ensuring that the proper defenses are in place. “But we need to go beyond that and make a difference in architecture and engineering,” he said.

NIST is contributing to this shift with its catalog of FISMA security controls, Special Publication 800-53 Rev. 4, which contains guidelines for agencies to specify trustworthy design and operation of systems being procured. NIST also is working with an interagency working group to develop guidelines for engineering and supply chain security, which are intended to create a foundation for a more secure infrastructure in the future.

And the agency is taking the lead in the creation of an Identity Ecosystem under the National Strategy for Trusted Identities in Cyberspace, which Jeremy Grant, head of the NSTIC National Program Office, calls a marketplace that will offer a variety of interoperable credentialing solutions.

NSTIC is intended to address the failings of current identity management and access control schemes. Although technologies to securely authenticate remote users exist, scaling them across large user bases and multiple applications is cumbersome. As a result, many users and applications default to less-than-secure schemes such as simple user names and passwords, which are vulnerable to a wide variety of attacks.

The marketplace already is responding to the problem, and companies including Google, Microsoft, Amazon and Apple have started offering customers multifactor authentication, Grant said. “NSTIC will provide a framework to facilitate interoperability between these solutions and others, and ensure that they are privacy-enhancing, secure, cost-effective and easy to use.”

“NIST is leading a public-private partnership to develop the Identity Ecosystem by funding pilot projects that are enabling consumers and service providers to obtain and make use of trusted credentialing solutions,” Grant said. “It is supporting creation of the Federal Cloud Credential Exchange to help federal agencies more easily accept trusted, FICAM-approved credentials for access to government applications.”

The NSTIC program office is working with the Identity Ecosystem Steering Group, a private sector-led organization formed to craft the legal, policy and standards framework to support the Identity Ecosystem.

Trustworthy systems and trusted identities will not eliminate threats, and administrators still will have to defend their systems from attacks. “We are at the critical crossroads in the next phase of the evolution of the Information Age,” Coviello said. “As we face an evolving and escalating threat landscape, it is clear that old, reactive, perimeter-based models of security are inadequate.”

Coviello is a proponent of using big data for this, harnessing data analytics for what he calls intelligence-driven security.

“An intelligence-driven security model consists of a thorough understanding of risk, the use of agile controls based on pattern recognition and predictive analytics to replace outdated controls, together with the ability to analyze vast streams of data to produce actionable information,” he said. “In an age of open, hyperconnected enterprises this is the only model that will allow us to handle known and even unknown threats, and to help reduce risk to acceptable levels. A model that allows us to detect attacks quickly and respond quickly, a model based on big data.”

None of these schemes to improve cybersecurity require creation of new technologies from scratch. But widely implementing and integrating them will be the work of the next five to 10 years.