Agent-based vs. agentless technology--do we have to choose?

We all know that a lot of IT departments are working under impossible resource limitations and time constraints, yet they’re still expected to deal with the latest security issues that threaten at every moment. So IT staff need the best tools to help them respond quickly and effectively to emerging threats without taking on any additional systems management hassles.

Agent-based solutions give a lot of in-depth systems visibility and control, but they also require a lot of resources and time to manage them. And while agentless platforms cost a lot less to deploy and manage, some IT professionals think they also offer less comprehensive functionality than agent-based solutions.

So what’s really best—agent-less or agent-based operations? Both bring a lot of positives to the table, so it all depends on your point of view:

Agentless pros:
•    No client footprint
•    True instant deployment
•    No provisioning requirements

Agent-based pros:
•    Allows total administrative control of machine
•    Ensures interoperability of 3rd party software
•    Gives ability to perform lower-level functions

These days, agentless solutions are more capable than ever, and newer “hybridized” solutions using both agentless and agent-based allow companies to reap the benefits of both. EMA wrote in a 2005 case study that “companies should consider altering their monitoring strategy to use an agentless solution throughout the enterprise supplemented by agent-based point solutions where appropriate.”

At KACE, we tend to agree that systems management is best served by a combination of agentless and agent-based technologies, so we built our KBOX Series 1000 appliances to take advantage of that. KBOX 1000 Series offers agent support for Windows, Mac, Solaris and Linux, manages agent and agentless nodes, and uses both managed virtual agents and agent-less network scanning for systems inventory. It brings the best of both worlds to mid-market companies.

Comments

[Trevor Smith]

The arguments against agentless have always been based around granularity and control. So arguably if these issues can be overcome one hase achieved nirvana. We at nanotechsoftware believe that we have achieved this with an agentless solution that can return data at the rate of 6,000 nodes per minute, with only a 1.5% hit on a 10 base network and the ability to monitor and interact with target machines; all without deployment of any code to the target machines

[Stephen Brown]

Agent vs. Agentless is always an interesting discussion. Denise Dubie of Network World recently published an article on the subject that might be of interest:
http://www.networkworld.com/newsletters/nsm/2007/0820nsm1.html

As far as which is better - agent or agentless - it really depends on the application and what the "agent" is...in the case of preventing private information from leaving the desktop it would be a piece of software, which provides very granular control but can be monumental to manage in an enterprise.

In the case of a network analyzer, this could be an appliance (probe) that is installed where ever granular visibility is needed. Here it's more a question of cost - not issues of management - you'll hear people talk about not needing probes because their collecting NetFlow or perhaps pulling SNMP information. But it comes down to the need for visibility and granularity.

You might want to check out my blog as you'll find we discuss similar topics - www.networkinstruments.wordpress.com