In-Depth
Forget Improvements -- Systems Management Needs a New Approach
Most IT organizations are only performing very basic monitoring on their servers. This new approach uses concepts we learned from the consumer gadget world.
By Denny LeCompte, VP of Product Management, and Jennifer Kuvlesky, Product Marketing Manager
Late last year, ESJ published an article by Travis Green that indicated gaps in application monitoring are caused by reduced staff and increases in change, resulting from greater adoption of virtualization. These gaps, he argues, will be solved with automation features in systems management tools. We at SolarWinds talk to hundreds of customers every day and have a very different argument regarding the challenges sys admins face and the approach to closing gaps in monitoring.
Monitoring Gaps Has More to Do with Organizational Barriers than Product Functionality
More than ever, sys admins are under a great deal of stress. With the increased adoption of virtualization and the ease of creating new servers as often as one likes, sys admins have a lot more to manage than before. Gaps in monitoring, however, are often caused by organizational issues. With virtualization, there is a new domain owner: the VMware admin or the cloud admin. There are also other domains including Windows/Linux Admin, Unix Admin, Database Admin, and so on. With the onset of shared resources (virtualization and cloud), domain roles are blending. For instance, we have seen that server administrators want to have visibility into the storage tier.
Inadequate monitoring results from domain owners choosing tools to manage their own domains without looking at the problem from the user’s or application’s point of view. For applications to work these days, monitoring needs to be covered across all of the various domains, including elements hosted in the cloud. Where we have seen this work effectively is in organizations where the line of business or application owner demands a comprehensive monitoring plan from IT.
Another challenge with centralized monitoring of composite applications, normally done by the IT Ops Monitoring team, is that the workers in charge of administering the monitoring platform are not subject matter experts in each tier of the application. As a result, they need to meet with each team, understand key performance metrics that should be monitored, learn what the thresholds should be for that application, and then use this information to set up the monitoring tool.
This brings us to our next point. Finding the time to set up monitoring for all of these domains is a very time-consuming task, and creating dashboards that visualize the aggregate performance of these application services is yet another daunting challenge that often requires professional services. On paper, having an end-to-end integrated view of the health and performance of composite applications is the dream of any line-of-business owner. The reality of implementing this vision, let alone the software and hardware costs to support it, is akin to going from having an architect draw up plans for your custom dream home to yelling at contractors nightly and spending the weekends painting the walls. It is hard, hard work.
Breaking Down Organizational Barriers with a New Approach to IT Management
How do we get the dream house built faster without cutting corners? First, IT organizations must seriously look at how they support the business and define the mission-critical applications that must be up and running 24 hours a day, seven days a week. What are the costs to the business of one hour of downtime in terms of lost revenue or lost employee productivity? What are the IT costs for lengthy troubleshooting procedures?
Next, you can find an IT management partner that does not impose barriers to IT management success. Some of the key requirements you should look for include:
Scalable, centralized monitoring that provides end-to-end coverage and domain guidance on what should be monitored and why. IT organizations don’t have the luxury of hiring an SME for each tier of the application. The monitoring tool needs to provide embedded smarts on what is important to manage for each domain. It should also be flexible and capable of managing custom or industry-specific applications to provide that end-to-end coverage sufficiently.
Visualization at the right level. The right level is the one that provides actionable information. Pretty charts that show abstract views of an entire network might look good in a demo to the boss, but they aren’t of much use on a daily basis. IT needs a UI that can provide one view of the health of all the components of an application, so when an app issue does present itself, the sys admin can quickly diagnose the problem to the failing component.
Easy to use, deploy, and manage. In practical terms, that means installing within hours, fully deploying within days, and managing the tool itself as little as possible. If the management tool requires professionals to install it, months to deploy, and a full-time equivalent to keep it running, it’s obviously not easy to use, regardless of what the vendor may claim.
Priced so that you can deploy end-to-end coverage in the same budget year. If you have to go all the way to your CIO to get the budget you need to deploy a monitoring tool, you’re spending too much. Powerful, usable tools can be found within an IT manager’s budget without requiring convocation of the entire IT hierarchy to move forward.
What is this new approach? The points above sound like basic, no-brainer common sense, right? The approach is new because it’s focused on the execution of a specific goal, keeping mission-critical applications up and running, but with the worker tasked to get that job done in mind.
This approach is much different from ITMAU (IT Management as Usual). ITMAU is focused on developing shiny features for the CxO -- features that will take a three-year roadmap to implement. The reality is that most IT organizations today are only doing very basic up-down monitoring on their servers and they spend too much time troubleshooting and reacting to application outages. The new approach to IT management is focused on moving organizations from reactive to proactive as quickly as possible.
This new approach is not revolutionary; it’s just an adaptation leveraging concepts we learned from the consumer gadget world: usable function that is easy to obtain.
Denny LeCompte is the vice president of product management at SolarWinds (www.solarwinds.com). Despite having a Ph.D. in Cognitive Psychology and spending some time as a psychology professor, he’s spent the last decade building IT management software, first as a usability engineer and then as a technical product manager. Denny believes that if a company understands its customers and the problems that they face every day, creating products that they can enjoy using becomes simple. You can contact him at [email protected].
Jennifer Kuvlesky is the product marketing manager for the Server and Application Management product offerings at SolarWinds. Like SolarWinds, she has roots in Oklahoma but has made her home in the high-tech capital of Texas for over 15 years, specializing in product management, strategy, and marketing with solid knowledge of the APM and virtualization management market segments. You can contact her at [email protected].