HP models security and trust
Pitching at security soup to nuts
Implementing sensible security measures is still one of those decisions that remains on the "what we should do next" list within many enterprises.
That is certainly the view of Martin Sadler, director of the Trusted Systems Lab at HP's research laboratories in Bristol. "Generally speaking there is still a low level of understanding about security and a lot of the devices that are accessing new services such as SOA (Service Oriented Architecture systems) and SaaS (Software as a Service systems) are out of date," he said.
His primary role of late, therefore, has been to build a security lifecycle model that can be implemented in any enterprise. The model his team has developed has this lack of understanding as its start point; getting users, particularly business users, to understand the importance of security risks is a key component in the success of implementing any security measures. This moves naturally on to the development of policy, the deployment of technology, securing the overall information systems infrastructure and monitoring the security and business processes as whole.
Security policy is, according to Sadler, one of the hotbeds of argument within enterprises. There is a lot of tension between the security personnel who want to beef up the technology deployed, and the business managements that do not want to make the capital investments that would necessarily follow. This process he calls Trust Economics.
To demonstrate the issues surrounding Trust Economics, the lab has built analytical modeling tools that can, like weather forecasting software, model business performance into the future and allow both sides to play "what if?" games with the parameters underpinning the business.
In this way, it becomes possible to analyse the tradeoffs between making extensive investments in security tools and implementing policies such as reducing security staff and/or investments. For example, the actual effects of events such as a serious virus infection of the IT infrastructure can be modelled and demonstrated.
One of the key technology deployments that can be made is in Identity Management - not only straight forward ID, but building maps of the relationship between an individual ID and that individual's role in the business. From analysing network data it is possible to map user IDs against access permissions and from that to map groups of users with similar permissions to roles that those users fulfill. Those roles can then be mapped to the roles that the business actually requires, so that role-based Access Control Lists (ACLs) can be developed.
These can be particularly important where a business has extensive internet-based relationships with partner companies. In such circumstances it is often important that individuals working for such companies have access to resources or services within the primary partner&'s IT systems. Allowing such access without comprehensive management controls is, however, also a significant security risk.
This process is also two-ended so that business managers can identify roles – and the permissions that they will require. This can then be used to map the access permissions associated with specific roles. "This can highlight access permission exceptions and anomalies where users have managed to acquire permissions that they should not have," Sadler observed.
He estimated that HP, with 50,000 partner-users, can save around $1m just by speeding up and simplifying the management of the ACLs.
Monitoring the business and security processes within any company often used to be an annual exercise geared to the yearly appearance of "the auditors". Now, security issues demand that it is a 24/7 exercise. This is only enhanced by the increasing strictures demanded by compliance to regulations and legislation. "The cries of pain over Sarbanes-Oxley compliance processing have actually been about the cost of fixing poor business processes," Sadler said.
To address this, the labs have developed analytical modeling tools that can help businesses identify the weak-points in both business and security processes that can cause the problems. One of the common causes is an historical adherence to meeting the annual appearance of the auditors, which means that post-audit, management, and control of business and security processes often slackens, only to be reigned in and brought under control when the auditors are next due to appear. By monitoring process activity as Key Performance Indicators the labs have been able to build tools that, for example, can be used to demonstrate to sysadmins when and where processes are being inadequately managed.
One final part of the security lifecycle model that HP has developed and is now starting to implement is the use of virtualisation technologies to overcome a common human issue that can have significant security implications. Many people work, at least part time, from home and run the risks associated with allowing a "work" PC to be also used occasionally by others.
Using virtualisation technologies, individual PCs can now be partitioned between different services, each with different levels of security, and with no "leakage" between partitions.
"So it becomes possible to have a single PC that has one virus-riddled virtual PC that's used by the kids, and one totally secure virtual PC for work, which operates fully under the control of the corporate IT environment," Sadler said.
This work is also being contributed by HP to the European Commission funded Open Trusted Computing consortium, which aims to build a comprehensive, open-source based trusted computing environment. ®