One such potential weak spot is around provisioning and deprovisioning of grid components, for example. The need to wipe sensitive data out of a component before it gets reused isnt unique to grid; its a concern anytime a system thats dedicated to one application gets recommissioned to work with another application.
But with grid, that can happen much more frequently, as components are virtualized and shared so as to increase availability and/or to maximize the use of resources.
The notion of a grid management entity—something that manages an entire collection of grid components— is another potential weak spot, as it presents a tantalizing target for DoS (denial of service) or access control attacks.
According to Lee Cooper, chairman of the EGAs Grid Security working group and Oracle Corp.s director of security program management, the EGAs Enterprise Grid Security Requirements presents enterprises with a list of security requirements that, although not mandatory, makes for a good checklist for organizations to compare against their own internal policies or risk profiles.
The documents requirements include, for example, making sure that all standard security between the management piece and the components of a grid setup is solid.
"Whether its encrypting grid between the management piece and the grid components or making sure data going across is good, thats a basic requirement," Cooper said.
As far as provisioning goes, the EGAs document urges users to make sure images on grid components, whether theyre of operating systems or security configurations, are trusted, Cooper said.
"With the provisioning process happening quite a bit, you want to make sure that images on grid components … are trusted images, that no ones mucked with those images," he said.
The EGA is a consortium of vendors and enterprise grid users that came together in April 2004, shortly after Oracle Corp. launched its 10g product line, with the purpose of driving enterprise adoption of grid.
Its Enterprise Grid Security Requirements document builds on the EGAs Reference Model.
That document, which the EGA released in May, contains a common lexicon of grid terms, a model that classifies the management and life cycles of the components required for enterprise grids, and a set of use cases that demonstrate the requirements for enterprise grid computing in specific enterprise scenarios.
After more than a year of hammering out security requirements, the EGA has reached the conclusion that enterprise grids are potentially more secure than traditional, siloed environments, Cooper said.
The rationale is that, for one thing, availability is easier to defend in the case of a DoS attack.
"If any component goes down, the application or service are more likely to be available, he said.
For another thing, a grid management entity centralizes security management, as well as the provisioning of components and user identities.
"You have one administration point," Cooper said. "Hopefully it will be more automated over time. You can manage security configurations across components, and youre in a better position to respond to various security events more rapidly than if you had several administrative points. And you are better able to manage provisioning of grid components, as well as identities."
Dan Kusnetzky, an analyst at IDC, said that security, as a subset of overall system management, is still a concern thats holding organizations back from adopting grid.
"There are many challenges to the adoption of distributed processing architectures such as grid," he said. "One of them is concern that application architecture can be picked apart, if its distributed across systems and those are available on public networks in some way."
Kusnetzky is in favor of standards in general, but stopped short of endorsing the EGAs security requirements as a cure-all, since the definition of grid and the market for it are simply too vast and diverse.
"There are six layers of technology in 21 individual software markets IDC is watching related to virtualization, and [the market totaled] $19.3 billion at the end of 2004," he said. "This is a very broad market … and a lot of organizations are using some component of it now."
The disparate array of technology mixes going to create whats being dubbed as grid technology can include, for example, linked storage systems, linked processing nodes, or virtual access environments that allow users to access applications regardless of where the application is located or hosted.
"Thats one challenge the market faces," he said. "Theres so many ways to set up a distributed solution. If the standard will be applicable to all is not clear."
Standards come in handy, he said, to minimize the staffing costs associated with managing these shifting beasts.
As it now stands, IDC estimates that software and hardware account for only 20 percent to 30 percent of costs associated with such technologies as grid.
Staffing takes the bulk, at between 50 percent to 70 percent of total cost over a five-year period.
"This is where standards come in," Kusnetzky said. "Increases in cost attributed to operation and administration can be minimized.