Chronosphere SaaS

I work as a data engineer, and we have many streaming pipelines. We use Chronosphere to monitor various metrics, such as how much data our pipeline is processing in each batch, the volume of incoming data, our consumption rates, and the time to process each batch. Additionally, we set alerts in Chronosphere for situations like job failures or when the number of processed records falls below a certain threshold. We get alerts if the record count drops below our threshold. Sometimes, we face silent failures, where our system appears to be working fine but isn't consuming any data because another system has stopped sending data. Chronosphere helps us detect these cases.

Another team member was involved in setting up a framework using Terraform on Chronosphere to monitor our job SLAs. We receive alerts on Slack or via email if any job fails to meet its SLA.

The alerting features are good because they provide many alerts you can't get with daily tools like Airflow or other orchestration tools, which notify you of job failures. With Chronosphere, I can set custom thresholds, such as the number of records being processed.

I can also configure alerts for silent failures and jobs that take longer than expected. Chronosphere is completely loosely coupled with your pipeline, meaning it has no dependencies on it, which is pretty cool.

It isn't very easy. It's not easy for everyone. It would be much easier if there could be a simpler version, like a data number version or an SQL version. It's hard to debug if you don't know the syntax. Also, I saw the Slack alerts feature, which is pretty cool, but I cannot customize my messages on the Slack alerts. It would be great if it were possible to tag people in the alerts. At DoorDash, we have hundreds of pipelines, and if something fails, I want to tag specific people so they can start working on the issue immediately. 

I have been using Chronosphere for one year. 

Sometimes, we noticed issues because we used the push method to send our metrics to Chronosphere using Prometheus.

I rate the solution's stability a nine out of ten.

100 users are using this solution, and we have a significant number of metrics coming in. It's highly scalable and can handle the load efficiently.

The documentation is pretty good.

If you want to monitor pipelines and use something like Kafka or any streaming platform, Chronosphere is the best option for monitoring pipelines with real-time alerts. It is loosely coupled with your pipeline, adding no confusion or load. I recommend using it.

Overall, I rate the solution a nine out of ten.

What do you like best about the product?

Our mid-sized startup was looking for more than our hyperscaler could provide, but we knew from experience that the big dog in this space would lead to expensive, painful tradeoffs. After (skeptically) evaluating Chronosphere I am happily convinced that there is a better option. My favorite things:

* Support: knowledgeable, friendly support from day one. I assumed this was pre-contract wooing, but one year later, support is as great as ever.
* Ergonomics: tools are only useful if they're used. Their interfaces load quickly and make sense, and as a result, our engineers are happy to use them.
* Operations: the product just works; the undifferentiated heavy lifting is handled behind the scenes, without incident.
* Cost: they include tools to identify and tame anomalous and low-value data, leading to lower costs without sacrificing signal.

What do you dislike about the product?

* Learning PromQL was a little painful. However, ChatGPT is great with PromQL, and that has smoothed the learning curve for new engineers.
* Prometheus histograms are clunky. It sounds like this may be addressed soon.

What problems is the product solving and how is that benefiting you?

Chronosphere provides our managed backend for observability (metrics, traces, dashboards, alerts). As compared with evaluated alternatives, Chronosphere provides more support, better ergonomics, and more powerful tools to control costs.

What do you like best about the product?

The ability to craft queries, ease of building dashboards and alerting

What do you dislike about the product?

The UI takes some time getting used to.

Additionally some tips for people new to using it like in AWS Cloudwatch insights would be helpful

What problems is the product solving and how is that benefiting you?

Observability and measuring performance of applications under varying loads

What do you like best about the product?

There are a lot of very helpful features such as muting alerts, easy to set up service accounts

What do you dislike about the product?

We recently had an outage where chronosphere wasnt reachable for 2h

What problems is the product solving and how is that benefiting you?

It's easy to integrate code with it to create features

What do you like best about the product?

Easy to setup and configure and easy migration of existing Prometheus resources.

What do you dislike about the product?

Nice to have cloud provider integrations.

What problems is the product solving and how is that benefiting you?

Monitoring and alerting on infrastructure and business-critical applications.