What needs improvement with Azure Monitor?

We cannot use AI services with the solution.

Establishing a threshold for resource utilization below 40% over a specified duration could trigger an automated scaling down, and vice versa. This way, the system could dynamically adjust configurations based on predefined conditions, such as scaling up or turning off resources. Enhancing and reaching a level of detail that facilitates pinpointing and addressing issues at such a refined level within the application and database components would be helpful.

I need connectivity with cost management.

Azure Monitor is not user-friendly, and the interface is not exciting. Switching between the dashboards is not easy.

I have used multiple products like Webex and PRTG. Some features could be added. Azure Monitor should add SMS and APIs. We have very limited access to Azure Monitor. I usually get alerts on my phone when they are integrated with Slack. I am not always available, but my team is. Sometimes, I am traveling and don't have access to my email, but I have Slack and other third-party projects that send me instant messages if a sensor goes down.

The solution's monitoring feature has limitations for analyzing multiple metrics. Also, its support services could be better.

In terms of pricing, Azure Monitor's billing based on data size can sometimes lead to increased costs, especially when developers need to purge data frequently. While there are mechanisms in place to track and manage this, there is room for improvement in terms of optimizing data pausing and related processes. Enhancements in this area could help mitigate potential billing concerns and provide a more seamless experience for users.

We encounter some difficulties in monitoring the operating system on its own. Therefore, we require additional tools to obtain a comprehensive view of the entire application chain. The technical support can be faster and has room for improvement. The dashboarding and reports could benefit from improvements, as Microsoft seems to prioritize Power BI as their main dashboarding tool. Perhaps these enhancements could be integrated natively into Azure, as there is certainly room for improvement in that area.

Its interface has room for improvement. Some of the options or some of the ways to navigate could be improved, but that's a general thing. All companies are always modifying the user interface because there's a better way to move things around or a better way to put things so that people can see the product features that they didn't know existed. For me, that's probably the biggest thing. It's a little lacking in features. There are some things that they could do to improve it, but it's not that big of a deal for me. There's obviously competition that does a lot more, but for what we're going to be using it for, it's perfectly fine. The biggest one is probably just the user interface. There could be more advanced logging at the database level. They can also improve their query builder to allow you to search for things better, but I last used it about a year ago. They might have already changed a ton of things in the newer versions.

They should include advanced logging on the database level in the Azure pool. Presently, we are not able to catch the transaction logs. The preview mode is available only for dedicated SQL servers and not for a database pool. They should improve this particular area of the solution. They should add a built-in feature for tracing the request and responses. Currently, it does not get measured automatically, and we need to write down the custom solution to check the request and the responses in the logs. They should create a log utility to log the request and send it to the Application Insights in the integration as a custom attribute. There should be default parameters and headers, which we can customize. It'll be beneficial because, at present, we only have responses, but not the requests. It is the solution's downside.

The query builder could be better. In comparison to other monitoring tools, in order to use Azure Monitor, your engineers need to have KQL experience. If they don't, it's not intuitive as a system. They need to understand KQL and get the right queries to get the value that they want, whereas a lot of out-of-the-box solutions, such as FrameFlow and Datadog, can be given to somebody untrained, and the UI will guide them through what they need to do. You lose some customization with that, but you don't need to train people on it. It would be good if Microsoft had some form of query builder in place so that you can choose a metric and it writes the code for you. Some kind of AI elements would help with that skill gap for organizations. Their support also needs to be improved. I've had a lot of issues with their support.

Azure Monitor's integration with applications could be improved. There is limited support in terms of what it can monitor, and the connectors are built in a way that only monitors a specific area. In the next release, Azure Monitor should improve its visibility features, especially for WordPress websites.

The APM needs to be improved to compare with Dynatrace or Elastic. For example, monitoring user sessions is interesting in Dynatrace or Elastic but is not dynamic in the solution. The static view is very basic. The monitoring of Kubernetes clusters needs improvement to be on par with competitors. With the solution, you need to monitor the process, deployment, application, and the security inside the cluster. This is not the case with other products. The solution should monitor or integrate with other cloud providers like AWS or DCP. That would be valuable because some customers have multi-cloud environments so they go with third parties to accommodate their needs. Because of this, customers say the solution is not interesting for them.

Alerts cannot be configured to monitor at a certain point in time. For example, we might want to alert people at zero hours but that is not possible. Splunk can accomplish this and its alerts are far better than the solution's options. The alerting mechanism is not up to the market. The default interface should be improved. You can prepare your own dashboard by using custom query language, but the default interface is not good.

Unfortunately, Azure Monitor stalls quite a bit. Azure Monitor can take up to 60 seconds to bring up metrics data. That length of latency is terrible and needs to be improved. The ripple effect of one wrong configuration affects multiple resources within milliseconds. Azure Monitor then reports after more than a minute that something went wrong. To improve this, Azure should create a visual representation of what the resource configuration was and compare it to what changed. Alerts are queries to figure out what has happened. If there was a reliable infrastructure diagram available, it could tell me where the configuration changed. Azure gives you so many logs, to understand where the change happens you have to review thousands of rows of logs. In the cloud, there are too many resources, so you end up trying to find the needle in the haystack to determine what is actually happening. In future releases, I would like to see Azure Monitor improve its diagram capabilities. Azure, in the last few years, has started to provide some basic diagramming where you can visualize from an Azure point of view, what is happening at the Kubernetes cluster and how the various resources are related to each other, we still need to use a lot of third-party tools. Imagine if an Excel sheet was thrown to you with a few thousand rows, and you were asked to determine what happened, within a minute or two, before a disaster strikes. A visualization tool is required to know what the previous configuration was as compared to the current configuration. The solution is also reactionary and not proactive or intuitive. Azure Monitor should be able to alert you that certain changes will cause certain outcomes before making the change using futuristic infrastructure diagrams. Lastly, I would like Azure Monitor to provide a separate portal for large operations teams, as there currently is no solution for them.

I would like more transparency when we use the solution with another environment, like on-premises, or on another cloud environment, like AWS or GCP. What I would like to see in the next release is not directly related to the solution, but it depends on Azure cost management. Maybe if we had the opportunity to monitor the cost of Azure and another cloud in one environment, and maybe integrate that with Azure Monitor, that would be incredible and amazing for any IT company to offer that to our customers.

If it is configured incorrectly, you can end up with a huge bill. Sometimes, a lot of data is collected and stored, and we may get charged for that kind of storage which is an issue.

We are sometimes confused about the details. We would like to have a flow in terms of how a new application should engage with Azure Monitor. We must understand: what is the most crucial part of our monitoring or support? We are not very good at it yet. Currently, it seems it's complicated to get the correct information in terms of what to do and how things work. Maybe it would help if they offered some guidance or written guides. For example, we say that project manager should follow their quality issues before putting the application into production. It's so complicated from their side for the project managers. Some kind of simplification is required for people engaged in the platform. Still, it's very new, and we will be gaining experience, and we may see much more substantial things in the future as we uncover more capabilities.

They need to work with other cloud providers - not just Azure. We'd like them to work with clouds such as GCP, AWS, and Alibaba, for example.

Azure Monitor is a complex product because you can use it to monitor different things, and that could be an area for improvement. Having templates by Microsoft added to it is another area for improvement. What I'd like to be added in the next release of Azure Monitor is an overview of best practices from Microsoft and what they'd recommend users to monitor. Assigning an alert is another feature I'd like to see in the next release of the solution. Currently, you can't assign an alert to anyone, if a server crashed and you need to assign that alert to a specific team, it's not possible.

Azure Monitor could improve the visualization aspect and integrate better with other third-party services.

Visualization in Azure Monitor needs improvement. Adding more features to it would also make the solution better.

If Azure Monitor wants to compete with other APM products in the industry, it has to stitch together the information and user flow. It also needs to provide all the information, as you get with Stock Trace, including details related to where the call came from and where the problem lies. That's what we'd like to see. With a solution like Dynatrace, you have a clear path and you know which calls were made before and after the problem, and where the issue occurred. That feature is lacking in Azure Monitor.

It's difficult to say as I only use a small part of this solution; however, It can be quite difficult to use at times. I think there are better solutions available. In my opinion, they should improve the overall user experience, especially when it comes to indexing and searching collective logs. The search language they've developed is quite difficult to understand.

There are a lot of things that take more time to do, such as charting, alerting, and correlation of data, and things like that. Azure Monitor doesn't tell you why something happened. It just tells you that it happened. It should also have some type of AI. Environments and applications are becoming more and more complex every day with hundreds or thousands of microservices. Therefore, having to do a lot of the stuff manually takes a lot of time, and on top of that, troubleshooting issues takes a lot of time. The traditional method of troubleshooting doesn't really work for or apply to this environment we're in. So, having an AI-based system and the ability to automate deployments of your monitoring and configurations makes it much easier.

Overall, we're not too enthusiastic about the solution. The solution is administration heavy, and very cumbersome. Using it is a lot of work on the admin side. It doesn't do things out of the box the way other products do. The new data ingest pricing model is wrong. They should scrap that all together. It should be a flat rate per year. As of right now, it fluctuates month to month, so it's hard to really know how much you will be charged. The solution needs a lot of work. It needs out of the box monitoring, real-time monitoring, and better network mapping. There are many features that are lacking. The process of implementation needs to be easier. The problem is the solution only does Azure. It can't monitor any other SAAS products. That's a very limiting design flaw.

The price could be lower but it is not a must.

I think the product could be improved if they were to work on automation related to gathering metrics from more applications without the need for instrumentation and things like that. I'd like them to work more on that. Additional features they could include would be more automations related to applications and monitoring, user experience monitoring on the application level. Improvement on technical support would also be helpful.

I'd like the solution to do more around vulnerability assessment. It's lacking in the product right now.

When something goes down, we want the option to have automation in place to get it back up again as quickly as possible. It is not very user-friendly so the GUI can be improved.

The troubleshooting logs need improvement. There should be some improvement there. I have a hard time finding the right logs at the right times whenever there is an issue occurring. They should simplify the logs.

I'm quite involved in the automation of dynamic alerts so I really can provide some insight into how that can be improved. They did have an alerts feature before and it was quite well automated, but since they've gone to a dynamic model and did a few updates, it broke a few things. Some commands that were used before no longer work. But some things that could be improved are: * Feedback should really be feeding back to MSFC (Windows Server Failover Cluster). * Make it so you can actually create more dynamic alerts on the fly. * Add capabilities to export templates dynamically. * Integrate better with Terraform — which everybody seems to be using these days. The integration with Terraform is broken on the Terraform side as well so it's not working as well as it should be from either point.

In comparison to New Relic, which I've used before, it's a bit more complicated. It's not as easy to use. It also took some time to get it working. The implementation needs to be simpler.

They can simplify the overall complexity since you have multiple data sources in the cloud for monitoring. It's quite simple, but there are so many portals. It takes time to work with it. If they could simplify the user configuration, that would be good. The integration capabilities could use some improvements.

They need to work on a more hybrid deployment that will allow us to monitor local on-premise deployments and connect to different systems. I would like to see more integration. In the next version, I would like to see more mobile applications. The current support is based on email so customers don't have full support. I think the developers need to put some effort into the support team to put skilled people in place and provide phone access.