Funny that you mention “An RMM tool is a very small portion of it – a commodity to anyone willing to pay a small amount of money.” in #10, above. Not more than a day ago did I write a blog post about exactly that point.

http://www.dedicatedit.com/blog/south-florida-computer-network-support-from-dedicatedit/reason-23-we-dont-rest-until-your-computer-network-is-safe/

Adam

Holding each RMM’s agent or agent-less implementation against questions like these should yield the best result for the MSP.

David Schafran
Transformation Strategies
http://www.TransStrat.com

Myth 1 – Agent based systems monitor more health check points than agentless solutions – this is key to ‘see’ more of the target managed system and to architect a highly resilient monitoring system

Fact – Agentless systems are policy based, meaning that they have a general capability to monitor any attribute via a monitoring protocol. That makes their capability unlimited when it comes to monitoring health check points. In contrast, most agent-based solutions feature built-in monitoring of selected attributes. Only a few support the ability to download policies for monitoring.

Myth 2 – Unlike agentless solutions, agent based systems continue alarm analysis and data collection on the local system when network connections are down, when connections are back up data is released to the central server – this results in no gaps in service level reporting.

Fact – A Windows server probably fails orders of magnitude more frequently than a Cisco switch. When a system is failing its agent is impaired and when it fails it is dead. Only agentless systems can monitor unavailability. When service is restored, agentless systems can continue to monitor seamlessly the transition from unavailability to availability. In the event an external connection is down, the local management software component (in Level Platforms case the Onsite Manager) can continue to collect data from all devices and maintain a continuous record of the entire health of the network.

Myth 3 – Agent based solutions function in an exception based fashion – meaning health check data is analyzed on the monitored system with only exception alarms being sent to the central management server, this eases the demand on network bandwidth.

Fact – Agentless systems operate in exactly the same way. In fact in our case the Onsite Manager also compresses the resulting data flow of exception conditions received from many end-systems, further reducing network traffic. This is something agent-based systems can never do.

Myth 4 – Agent based solutions are typically much more robust for event log monitoring.

Fact – Agentless systems can capture any event with the added flexibility of policy based management to easily add new events. Agentless technologies have the additional ability to catch event log entries made by the operating system between the time when all services have failed (including any agents) and when the system has restarted those services.

Myth 5 – Regarding agentless monitoring being quick and easy when compared to agent-based solutions, I would argue against that notion… agentless solutions require monitored systems to have proper remote access services and protocols available and running on the target systems (i.e. WMI, SNMP, SSH, Telnet, etc.) – in many cases these requisite components are not running or missing and the effort to get each system ready for agentless monitoring can equal, or more likely exceed, the time and effort to install agents across numerous systems. Let’s also consider firewall configurations for each agentless monitored system – this is a significant effort as well.

Fact – Ensuring that end-systems that are properly configured is a basic requirement for anyone offering managed services. Agentless systems enforce this fundamental best practice. As these technologies are more widely adopted the vendors are moving quickly to automate these processes so that any disadvantage is not only minor but temporary. The ability of an agentless system to immediately detect and manage new assets is a major advantage.

Myth 6 – Adding to the potential challenge of setting up an agentless environment are its limitations in monitoring and architecture that could introduce risk. Many agentless monitoring solutions lack resiliency; unlike agent-based solutions, with an agentless approach there is a risk of lost data in instances of broken connections between the central management server and target system resulting in gaps in service level reporting

Fact – This was addressed above (see Myth 2).

Myth 7 – Agentless is usually ok for basic operating system and network monitoring but is typically less able for deep network, database, and commercial/custom application monitoring.

Fact – Deep network monitoring is actually easier with agentless because it inherently requires consolidating a view across many network elements – none of which can run agents! Deep monitoring of databases is better handled remotely because they all support a remote SQL capability in order to support client/server applications The same facilities can be employed to provide deep monitoring of database system and application tables. The resource consumption of all applications can be monitored deeply via remote monitoring of system performance counters. Custom applications that follow best practices are also readily monitored remotely because they write events to the Windows application event and security logs. Even applications that write to custom logs can also be monitored remotely via WMI. In fact with the advent of virtualization, “deep monitoring” is actually much harder to do with agents since they can never monitor more deeply than that virtual machine that they run within. Agentless solutions can also monitor the hypervisor layer that is hidden from the agent’s VM viewpoint.

Myth 8 – Log monitoring (ascii, syslog, event) may be lacking and inefficient (i.e. transporting log file contents across the network for central processing) or non-existent.

Fact – The majority of standard protocols such as WMI, WS-Man, SNMP/Trap provide a way to filter content. It is not necessary to transport entire log files.

Myth 9 – Agentless monitoring consumes excessive network bandwidth due to frequent communications and heavy data collection between the central management server and abundant target systems.
Fact – In our experience there is no measureable impact on local network performance and significant efficiencies in communicating externally to a remote service management center by centralizing and consolidating external network communication.

Myth 10 – Agentless system monitoring is typically limited in the volume of unique health check points that can be observed on the target managed system, this limitation introduces risk.

Fact – There is no counter in a Windows system that cannot be monitored remotely.

Myth 11 – There certainly are good reasons for agentless monitoring, for example;
- Where deep system/server resident monitoring is not a requirement
- When lack of monitoring robustness and resiliency does not introduce risk
- When agents are not allowed to run resident on the customers servers
- Monitoring application performance from the end-users perspective, i.e. transaction simulation

Fact – None of these conditions need to be present to justify agentless technology’s advantages.

Nimsoft Concluding Observation – 10 of 10 Nimsoft customers choose agents or a hybrid approach.

Fact – While I am sure this is true, I suspect this has more to do with Nimsoft than their customers. The hybrid approach which virtually all the agent based systems are using to deal with the growing inadequacy of agents represents the worst of both worlds. In a hybrid environment, the MSP must keep track of (a) which systems have no agents and are not being monitored (b) which systems have agents (c) which systems have no agents and are being monitored agentless and (d) which systems have agents that also provide agentless monitoring of other systems. Consider the cost of administering this complexity compared to the simplicity of the agentless solution where none of the systems have agents and all are being monitored through one consistent set of centrally established policies.

Peter Sandiford
CEO
Level Platforms
psandiford@levelplatforms.com

Level Platforms new Offsite Assistant is in beta today and will be released shortly as part of our Mnaged Workplace 2009 r2 release. This is a network-based solution and does not require the deployment of a management agent on the device, thereby maintaining all the benefits of agentless systems without introducing the significant problems and limitations associated with agent-based systems. Anyone interested in learning more about this breakthrough technology is welcome to contact me directly.

Peter Sandiford
CEO
Level Platforms
psandiford@levelplatforms.com

The analogy used of the sculptor and his tools is quite appropriate. If industrial designers today were limited to “hammers and chisels” I am not certain what our world of steel and plastics would look like today. I am not sure any of us would want a hand chiseled automobile no matter how skilled the workman and his team.

My point is simply that cloud computing, virtualization, SaaS, unified communications, etc, are the technologies of tomorrow. Client server computing and the companion agent based management “tools” designed for this era are hard pressed to support these new technologies. Emerging standards across the computing industry over the last decade are redefining the tools required around an agentless future.

MSPs can be successful with both agent and agentless technologies today where we continue to frame the problem in terms of PCs and servers. My argument is simply that the world of network-based computing is already moving onto center stage. MSPs with the appropriate tools will have a growing and ultimately unassailable advantage as they apply their ingenuity and business skills to addressing their customers’ needs with the tools appropriate to the task.

Peter Sandiford
CEO
Level Platforms
psandiford@levelplatforms.com

Anit Chaudhary
Manager
Real Time Data Services

Now agent vs agentless, vpn vs. no vpn. I have had the opportunity to support multiple clients of every imaginable configuration utilizing a number of tools mentioned, and not mentioned. IMO, thus far, agent based wins out over agentless if you account for all the time spent deploying and managing the tool and it’s idiosyncrasies, and really, that is what the tools you manage your customers with are for, to save your time and your customers time. Now, that being said, we are in the process of evaluating one of the agentless products and run it thru it’s paces to see what if it can deliver what we require. We’ve previously used two (mentioned in the comments) of the agentless companies, but it’s been years. Back then, their tools were definately not up to the challenge and we replaced them with agent based tools. Today the agentless architectures available might be mature enough to work as required.
We also maintain static vpn’s to all clients. We don’t monitor over them, but use these ease our remote access to certain devices and securely deliver all of our services. All of our tools will run securely without the vpn, but having the vpn in place allows us to provide more rapid problem resolution in most cases. We’ll have to adjust as more SaaS offerings emerge, but thus far it works well for us.

Of course, included in the list of proponents for one method versus the other are representatives (like me) of the vendors that provide their solutions and guess what…they will create all the arguments in the world to support the capabilities of their products.

Customers and therefore MSPs must have flexibility. They must be able to choose between agent based or agent-less or, in many cases a combination approach. Nimsoft offers both, so, just like the other vendor CEOs, I am also supporting our company position. This is why Nimsoft tends to be chosen by the more established MSPs who have matured beyond the limited approach of some of the lower end products (that are very good by the way for smaller MSPs getting started).

Nimsoft has recently replaced Silverback at a number of high profile MSPs, including one who wanted to use agent-less and the other wanted agents. Interestingly, the agent-less MSP “absolutely refused” to use WMI as the mechanism for managing Windows servers. The extract from the customer’s email to us…. “The reason this would not work for us currently is that we have approximately 1200 windows systems under management today, and WMI is very problematic to say the least on the newer windows based platforms”. This is a real life MSP, dealing with real life management issues – not a vendor supporting their position.

Nimsoft responded by adding in other methods for getting the data from the Windows server platforms (RPC).

One final comment…agent-less for Cloud? Resources that are there one minute and gone the next, that don’t have a fixed IP address, that communicate their information across the public internet, that you need secure, encrypted and guaranteed delivery of the monitoring information. From our research to date, and we have both Cloud providers and Cloud consumers using Nimsoft, it looks like an agent-based approach is definitely the preferred method but we’ll see….having both capabilities means that we can take a neutral position.

So…don’t believe the vendors who arguing one method to the detriment of the other simply to position themselves correctly.

Bottom line – as an MSP you want flexibility – ensure that any solution you invest in has the flexibility and breadth that you need.

Gary Read
CEO
Nimsoft
garyread@nimsoft.com

Akash Saraf
CEO
Zenith Infotech