Skip navigation

Geek Speak

4 Posts authored by: aluciani

Mobile computing is pervasive, with more and more work done on mobile devices.  Folks may access your website through a mobile browser or native mobile application, does this change the entire web performance monitoring stack or can we use the same applications to monitor mobile and web apps?

 

There seems to be a lot of discussion around mobile application performance monitoring, which provides an end-to-end stack visibility, all the way from the device to the back-end services and infrastructure, and real-time visibility into end user experience.  With a large variety of devices in the wild, how do you measure end user experience and ensure that your mobile services are working well vs your website.  The metrics that you report on will be different vs standard website performance motioning. In fact, you get more insights into how users use your web services being able to track deeper into the user experience on mobile.

 

Mobile applications at times offer more complexity, since your dealing with different devices, software versions, network converge and latency.  Its imperative you have insight into how these areas are performing in addition to your website.   Incorporating crash reporting into your mobile devices provides you with an additional level of intelligence, that you likely wont get from your website alone.

 

Some folks see mobile application monitoring as another level up the stack, but I see it as on par with web application monitoring, in terms of getting metrics on user experience.  I also believe that mobile application performance monitoring is a set of tools used in addition to your web performance monitoring, but in some cases may be the same product.

 

Are you measuring end-user experience on mobile devices?

 

Do you feel mobile application performance monitoring is a different discipline than web performance monitoring?

 

Do you correlate issues between your mobile web applications and your website?

 

Are you transitioning more of your web applications to mobile and how are you monitoring their performance?

 

Do you feel mobile application performance monitoring is a new area of growth or just a fad?

I am a big fan of data and metrics (.. that is data that we can turn into knowledge and Commander Data from Star Trek). In IT we tend to collect a lot of it, especially as it relates to our infrastructure and reporting it within to our business folks is sometimes a challenge. For one, we tend to report in terms we understand, take for example the following DNS resolution time, TCP connection time, HTTP redirect time and Full page object load time. But do business folks really understand what we are saying?

 

Its important that our systems and reporting translates those metrics into understandable business language, an intuitive web dashboard that helps both IT and Business understand the impact of a web performance metric. Take for instance a simple dashboard that has a green - yellow - red system, for indicating problems and the metric next to it is a roll-up of different IT data points.  Its some much easier to deal with - Web forums from Austin OK rather than process xyz on server-abc is in a not running.  Its easier because you know the complete impact of the issue rather than just focusing on one issue and not understanding how its tied to other items.

 

I think machine learning is going to play an important part in future web performance monitoring applications, as business demands more from IT.  Machine learning will help correlate IT and business metrics. Business will want to know more than just if something is down or up, they need metrics delivered in a way that help them understand business impact.  BTW - Commander Data would also communicate in terms the Captain and staff would understand after compiling the data.

 

Are you more concerned about actionable IT metrics and not concerned with the business impact?

 

What are the metrics that matter to you?

 

Do you translate IT metrics for the business or do they manage a different set of metrics?

 

Have your business teams stared asking for more relationship mapping between metrics?

 

Do you see machine learning as a next phase in matching IT and business metrics?

As more software is delivered through the web as software-as-a-service (SaaS), web performance monitoring tools are no different.  There are a number of companies that allow you to pay a monthly fee to use their service, but many people feel that, it's better to manage your infrastructure or web application using tools in-house.  I think there is space for a hybrid option, which can take advantage of both scenarios.

 

One of the benefits I see with an external monitoring tool, is you get to see the performance of your web application from an internet user perspective.  I've seen issues when externally, the application is performing poorly but internally folks don't have the exact same experience.  Then the old tried and true call comes "... can someone use the cable modem and test out our site".   Depending how deep you want the external monitoring to go, there are times when you need to install an agent on each server you want managed and that can add some overhead or not and your security team may not be too thrilled about that.

 

Some folks, prefer internal monitoring only, but I don't think they are getting the complete picture when it comes to external web performance.  Internal monitoring can at times alert you to issues before they become know to the broader community.  Some folks have also found that moving monitoring internally that they are not dealing with issue outside that may skew their performance results, whether its a poorly operating content delivery network (CDN) or bad internet connection from their provider.

 

What are your thoughts on in-house vs hosted web performance monitoring solutions?

 

Is there are difference between the two?

 

Do you leverage both services, targeting specific areas for each to monitor?

 

Do you prefer on over the other?

 

Does cost come into play when deciding on a solution?

 

Let me know your thoughts, and how we can build better monitoring services that can manage the entire stack from infrastructure to application.

Imagine starting your day in your “war room” with folks from different parts of the IT organization.  Your teams have been assembled to deal with the typical “Our website is slow…” comments from external customers and internal business units.  You have a multi-tiered web application with the latest hardware and application code.  As you go around the table, the first response you typically get is “.. not my problem, maybe its…”.  Where do you start?

 

When dealing with web performance monitoring, folks typically start at the infrastructure or application layer, but this siloed approach can only get you so far.  Multi-tier applications are interconnected and very complex.  We need an end-to-end approach that will help connect the dots.  This approach could also help prevent unnecessary hardware spend, a typical response from folks who think just adding more compute, memory or storage would solve the issue.

 

I would be interested in hearing from you regarding which tools you use to help monitor your web performance.  There are a number of products that provide a holistic point of view that could help to reduce your Mean Time To Repair (MTTR).   Solutions such as Pingdom, Keynote and AppNeta may be the tools for you.

 

Do you currently monitor your website performance?

 

Do your users notify you when your website is slow?

 

Do you have a passive or active monitoring solution?

 

What kinds of features would you like to see from current day solutions?

 

Do you use a hosted or in-house web performance monitoring solution?

 

We have had a lot of success with web performance monitoring tools, let me know how things are going for you.

Filter Blog

By date: By tag: