VMware ESX Server, the enterprise-level virtualization product from VMware, is the leading virtualization technology in the world, as everyone knows. This hypervisor-based architecture (called the “VMkernel”) works in coordination with a Linux 2.x kernel, directly driving hardware configuration to configure VMs. VMware ESX includes the Linux-based Service Console which is used as the bootstrap for the VMkernel, and secondarily as its primary management interface.

Many ISVs who specialize in agent-based hardware and system monitoring have until now found it fast and easy to support the VMware ESX environment by just deploying their agent inside the Service Console (and in some cases inside the Virtual Center and in each VM as well). Some experts raised the red flag that the Service Console can be exploited as a security breach because it relies on Linux security. Partly as a result, VMware began recommending in mid-2006 that ISVs (and customers) avoid running third-party software in the VMware ESX Service Console.

But the real kicker came in September 2007 when VMware announced VMware ESX Server 3i, which eliminates the possibility of deploying any software component within the Service Console. The VMware announcement states: “ESX Server 3i completes an ongoing trend of migrating management functionality from the local Service Console to remote management tools,” and describes in a table that the only way to manage and monitor the ESX 3i environment is with agent-free technologies (known as agentless data collection). From VMware’s perspective, they’ve strengthened security, improved reliability and simplified the management environment. For members of VMware’s ISV community who are invested in agent-based approaches, this means they’ll have to move very quickly to transform their product architecture. Within the VMware ecosystem, this has rippling impact.

But with the right approach to virtual machine management, running or not running an agent inside the Service Console is just a sideline issue. The real objective should be to have an architecture that collects data remotely but computes and processes centrally. In most agent-based systems, the agent acts as the “brain,” with the function of the centralized component limited to storing data and providing a web portal. From an R&D perspective, moving from an agent-based to agentless architecture is costly and may scare some vendors. If the trend of VMware’s September announcement continues the largest ISVs facing this conundrum may think about making acquisitions; the smaller ones may disappear.

InfoVista’s server management customers don’t need to worry about whether their investments are sound. We’ve taken an agentless approach for 12 years — and also offer embedded support for many existing vendor agents. We — and our customers — are fortunate that our R&D budget can be invested in supporting new technologies and capabilities rather than having to revamp our core product architecture.

0 comments so far...