Steve Francis (Global) - The Top 5 Reasons Monitoring Sucks

• 0

• Posted by Steve Francis

• on June 13 2012

There’s been some interesting discussion around “Monitoring Sucks”. Go check the Twitter hashtag #monitoringsucks to commiserate. This isn’t a new opinion – it was because monitoring sucked that I started LogicMonitor. As technologies get more complex and companies deploy multiple datacenters across topologies – public cloud, private cloud, hybrid cloud – monitoring becomes more challenging, downtime becomes more likely, and IT professionals become more hateful of monitoring tools.

Top 5 reasons why people think monitoring sucks:

#1 – Monitoring tools expect me to be a monitoring expert

Almost any monitoring tool will collect SNMP data, and so claim they can monitor any SNMP device. True, but the magic is in knowing what to monitor. A tool may have profiles to monitor common Cisco switches, but if I roll out the latest Dell Force10 switch, it’s up to the admin to figure which of the hundreds of SNMP objects that could be monitored actually matter. Tomorrow, when I add a new NetApp storage volume, I must remember to add it into monitoring.  Worse, if I upgrade NetApp software, I have to know they change the units for latency from one version to the next.

Suck Remedy: A monitoring solution with embedded intelligence that “knows” what to monitor in your infrastructure, sets default thresholds, and automatically identifies changes to existing devices and new devices.

#2 - Associating metrics across my infrastructure is too difficult!

Most monitoring systems collect megatons of data - it’s making sense of it that’s difficult. If I have a pool of 50 front-end web servers that should all get a similar load, I want to know if there are outliers without looking at 50 graphs.  To understand the efficiency of my application across various usage patterns, I want to graph ratios, not raw data. I want to compare my infrastructure hosted at Amazon with my Co-lo in Chicago.

Suck Remedy: A monitoring solution that works across locations and easily creates custom views of operational intelligence – such as showing the top 10 busiest servers from a set. And easily plots the arbitrary ratios given above, for a varying set of machines.

#3 – Monitoring is a time suck!

It can take weeks to properly set up and deploy open source monitoring tools, and many commercial toolsets require lengthy professional services engagements. These solutions are powerful tools in the right environment – but require highly skilled engineers.  Even managed service providers are dependent on senior engineers to manage monitoring. This creates a monitoring “gatekeeper” culture.

Suck Remedy: Monitoring with quick, automated setup that includes embedded intelligence for hundreds of devices, with built in thresholds, so you don’t spend your time determining what and how to monitor every metric on a device. Also, an easy-to-use interface accessible everywhere and to everyone in your organization.

#4 – Repetitive collection of metrics drives me nuts!

Parallel systems that collect the same data multiple times - once for alerting, once for trending, reporting, etc. - is inefficiency and frustration defined.

Suck Remedy: Consolidate all data collection, all monitoring – into a single solution.

#5 – I have too many monitoring tools!

Amazon monitoring for cloud deployments, a premise-based solution for the corporate datacenter, Mongo’s MMS for the new database, and specialized storage monitoring tools. The patchwork approach might work, but sacrifices time, resources and sanity. How many companies have the combined skill set necessary to manage and deploy so many monitoring tools? Not to mention the many downsides of creating information silos and the hassle of managing alert escalations across multiple tools.

Suck Remedy: A single monitoring tool with consolidated alerts, unified data collection, and infrastructure-wide performance intelligence and operational intelligence.

By monitoring your entire infrastructure in a single pane, you can alleviate multiple pains, and ensure actionable performance intelligence, operational intelligence, and even real-time business metrics that your C-levels are going to love, like revenue trends, sign-ups, and active users. This is the flipside of monitoringsucks, which is what, #monitoringrocks?

By Steve Francis, founder and CPO of LogicMonitor