Open Q&A
If there's no good subforum for your question - ask it here!
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Transitioning from Multidimensional ?

rgarzon1
Champion

Hi  

I’ve been working with Dynatrace for a long time, and Multidimensional Analysis has always been one of the most powerful and straightforward tools in the platform. Getting the p99 response time for Database statements (split by {Request:Name}) used to take just a couple of clicks.

Now, with the push towards the new Distributed Tracing app, what used to be a routine daily task has become an incredibly frustrating experience. Beyond just losing the intuitive UI for custom percentiles, I am running into several critical issues:

1. Broken Migration Workflows When I use the UI prompt in Multidimensional Analysis to "Switch to the Distributed Tracing app", it simply breaks the tracing. The context is lost, the feature doesn't work as expected, and I have to start from scratch.

2. Unexpected Consumption Costs & Incomplete Data Because the UI no longer gives me the p99 aggregations easily, I am forced to query the spans directly. This approach has already generated  consumption costs just trying to replicate a basic, previously free dashboard view. To make matters worse, despite the cost, the data returned feels incomplete compared to the old analysis.

rgarzon1_0-1787124131849.png

 

3. Overall QA and Broken Links in New Tenants Honestly, it feels like there is a lack of testing behind these UX changes. I am finding multiple broken links across the new tenants, which gives the impression that the user journey isn't being thoroughly tested before deployment.

Am I missing a straightforward, cost-effective workflow inside the new UI to get my custom percentiles back? Is anyone else struggling with these broken links and unexpected query billing?

The banner on top says: "Switch to the Distributed Tracing app - replacing Multidimensional analysis; or use Notebooks for advanced exploration". But if basic, out-of-the-box analysis now requires writing paid DQL queries in Notebooks just to see a p99 on SQL queries, this feels less like an upgrade and more like a workflow downgrade with an added cost. and not even its the same result 

rgarzon1_1-1787124317245.pngrgarzon1_2-1787124381610.png

I would love to hear how other teams are handling this, and hopefully, someone from Product Management can shed some light on these QA and usability issues.

fuelled by coffee and curiosity.  searching for a job,
3 REPLIES 3

Julius_Loman
DynaMight Legend
DynaMight Legend

Could not agree more. There are definitely gaps between MDA and current Gen3 apps, and the only way is to use DQL span queries.

Services app provides some of the performance analysis, but lacks the custom percentiles or even basic percentiles on non-request queries. CC @JohannesBraeuer 

Percentiles in Requests

vs database queries or outbound calls where percentiles of response times are essential:

Missing percentiles in DB queries

 

The distributed tracing app overview screen does not allow specifying custom percentiles either and does not include p99 CC @tom_rothschaedl . 

Missing p95 and p99

Missing p95 and p99

 



Dynatrace Ambassador | Alanata a.s., Slovakia, Dynatrace Master Partner

dannemca
DynaMight Guru
DynaMight Guru

And don't even try to create metrics from spans to try save licensing... You will get samples only, and no percentile available... 😓

Site Reliability Engineer @ Kyndryl

nick-p-montana
Visitor

The consumption costs for span queries on Dynatrace Grail are absurd especially for large environments. I think customers should try to utilize bizevents as much as possible simply because they are less likely to decimate limited budgets

Featured Posts