Dynatrace Classic

My main use case for Dynatrace is observability, monitoring, insights, performance, variation, dashboarding, services, traces, validation, cross-team collaboration, reporting, and many more.

As a senior performance analyst, I look into Dynatrace in terms of the observability parts, and mostly, there are multiple dashboards that have been created for multiple product groups from front-end layer to a back-end layer. I keep monitoring their performances, their failure, and their anomalies that have been detected through the problems page, alerting if it has been sent to the right team, and investigation on why these issues do occur, how are they recovered, and who would be the right team to contact. One of the newest cases would be when we had extremely high load on our application, and we were unable to figure out what was causing it. We looked into Dynatrace and noticed Apdex being down, high load days, and request errors being increased, and investigated to identify that it was the bots that got increased, but they were all coming as real users. We tried to investigate their user agents, we tried to investigate their IPs, connected through a bot manager system and CDN, and then mitigated the issue through content protector. We were able to improve our performance, our Apdex, and we did reduce our overall load on the site.

In terms of my main use case and other ways my team uses Dynatrace, I would say teams have collaborated. DevOps uses it for administrative purposes, including me, and then we have developers use it for investigative purposes on our lower environments like QA and Dev. We have a separate tenant specifically only for China users because they are restricted through firewalls and all of those things, so the servers are all set up within China, Tencent, providing a specific tenant only for them. These are most cases.

Dynatrace has impacted my organization positively in many ways because we have been using it for almost eight to nine years now. We have used Dynatrace day in and day out for performance improvements. The recent example I would share is from our front-end team, which used it in daily investigations of how our web vitals, including Google Web Vitals, are performing in terms of LCP, INP, DOM interactive, Time to First Byte, and other web vital metrics. We were able to figure out that some of our images or some of our JS scripts were the primary cause of these being down, and we were able to use Cloud and other agents through Dynatrace to find the slowest performing components, fix them, and we have been able to see those improvements. Our site is way better compared to three months ago.

To measure those improvements, we see that our LCP actually improved by about 42%, trending around 5.3 seconds on a median range, which we improved to around three seconds range. In the same phase, we also improved our TTFB by somewhere around 40-42% and DOM interactive by somewhere around 40-45%. We are also trying to improve on multiple other categories, not only on overall site performance but probably homepage performance, content performance, product page performance, search page performance, and others. These improvements are in progress.

The best features Dynatrace offers are many. I would say from dashboarding, alerting, integration with multiple applications, the new DQL features that come up with Dynatrace, the notebook feature, the workflows, the ready-made dashboard that Dynatrace provides, the page where you are able to track your license consumption, and the observability part where it provides the root cause for the incidents that appear.

Out of all these features, dashboarding has been a major contribution in my day-to-day work, looking at performance trends, failures, errors, and all of those things, and cross-verifying that with the problem detection that Dynatrace provides. Over the recent past, DQL has been a major help, especially for dashboarding purposes, investigative purposes, particularly through notebooks, since you do not have to run a dashboard always; you can just use very less license consumption and run a query to find out what the actual trend looks like or what the failure looks like and where exactly this is. Most of these are collaborated with Dynatrace Assist, AI Assist, and Cloud Assist through MCP servers, which has also been pretty helpful.

Certain configurations still require some navigation through multiple screens. Dynatrace offers many apps, but when it comes to training and documents or help from a point of view, it is very elaborative. You need to go find, search, look into the specific ones, and see how they are changing. Dynatrace currently offers both classic and Gen3 versions, so most of these documents are kind of updated with Gen3 while some users may be using the classic version, making these documents sometimes confusing regarding their availability for classic or Gen3 versions.

I would also like to add that the user interface has been helpful, allowing all the apps it connects to be accessible through a single stop or a button, where you can get all the apps and then look into them. So in terms of user interface, it is pretty helpful.

Service capabilities are great, user communities are great, the relationship with Dynatrace support, CSMs, or account managers has been great, consulting partnership is amazing, and the breadth of services is awesome.

Regarding Dynatrace's AI capabilities, I think its governance and security are handled well because until this date, there have been no complaints or issues we have come across. I agree that the governance has been taken care of, and all these CCNA, CCPA, and all PII level priorities are considered, so they have been good in that respect.

Regarding the accuracy and reliability of output of Dynatrace's AI capabilities, DTCTL has been great, and I would probably give it a percentage of somewhere around 80-85% on its accuracy. Through Dynatrace Assist, although it uses agent AI, there are times I have seen not satisfactory responses. So I would probably give it somewhere around 75%.

For others looking into using Dynatrace, all those examples I shared earlier in terms of pros have been great. I would say anyone really willing to purchase Dynatrace can use their test features for testing or POCs that Dynatrace would provide for approximately a month period. This would be helpful in understanding what the tool can provide and how beneficial it is for your organization. I rate Dynatrace a nine on a scale of one to ten.