Introducing unlimited request plans

We are introducing new plans for Tideways and are retiring the request-based pricing on March 15th 2017. You can already visit the new pricing page with all the information.

New plans are billed per application and include unlimited monitored requests from unlimited servers. Tideways is primarily an “Always On” production profiler and tracing system and therefore pricing is now heavily influenced by how many traces we collect every minute and how long we store them for you. With the new plans, you decide on an upper limit of detailed traces collected from an application every minute, for example 25 traces/minute – which translates to roughly 1 million traces a month. Each new plan has a number of included traces/minute, but you can also increase the traces/minute limit in steps of 25. With this model, you have guaranteed fixed monthly prices you know in advance, based on your desired level of traces/minute, but independent of application size, architecture and host count.

If you are using Tideways as a traditional Profiler to manually generate Callgraph profiles, you will also be happy to hear that we increased the retention for manually triggered traces to 30 days, regardless of the plan. In the Lite plan this includes 10.000, in the Basic plan 25.000 and in the Standard plan 50.000 triggered callgraph traces per month. You can also purchase a dedicated upgrade to increase the retention to 60 days and add 250.000 additional callgraph traces/month.

For existing customers we are grandfathering all current plans and give you the choice to stay in your current plan or upgrade to the new plans if they suit you better.

What are the new prices? A Lite application with limited APM functionality and 24 hours retention costs 39€/application/month, the full APM functionality is available in either a Basic application with 3 days of retention starting at 89€/application/month or Standard application with 14 days of retention starting at 159€/application/month.

What is my benefit switching to the new unlimited request plans?

Our goal with these new plans is to make it simpler for you as a customer to calculate your Tideways costs up front and independent from month to month changes caused by server up- and down-scaling, seasonal business and application architecture. You decide on the level of visibility you want into your application and the duration we store this data for you and we optimize everything else. Instead of monitoring costs that directly scale with your application request volume, you can now decide yourself when and why to increase the amount of traces/minute because your application has grown in code-size and functionality, production traffic or both.

This is the next step away from the per-server pricing that many of our monitoring competitors sell. We think both servers and requests are a bad proxy for the value that you or your customers get from monitoring a particular application. Especially in this age of multi-cores, microservices, Docker and vertically scaled applications for increased reliability. You shouldn’t be punished for your choice of deployment technology, host topology or application architecture. With the new plans we can offer fair prices to customers that run on a single server and want to keep hundreds of traces every minute and customers that run on 10+ hosts but are fine with only a very limited visibility.

When you are upgrading to the new plans the minimal traces/minute configuration of 25 traces/minute for a basic plan is higher than the previous limit of 20 traces/minute that we built into the old plans.

I am not convinced, can I keep my old plan?

Yes, you can! We are grandfathering your current plan. You can continue to upgrade within tiers of the old per request model. You cannot subscribe to the old plans if you are not yet a customer. We do reserve the right to provide new features for the new plans only.

Remember that the new plans are designed to increase the visibility in your application, they store a higher amount of traces every minute than Tideways did before. Upgrading will give you access to more traces and better insights.

How does this affect Microservice/SoA architectures with multiple applications in Tideways?

We have released a new feature called “Services” last month, which allows you to collect data for a microservice/SOA-architecture into one application in Tideways and view performance statistics for each service individually. We suggest that you migrate your deployment to using services instead of multiple applications. Since this feature is only available to Medium (M) plans, if you have this problem and are subscribed to a Small (S) or Lite plan, please contact [email protected] and we activate the feature for you to prepare for a plan upgrade.

I have many tiny applications and the new pricing per application is bad for me

Unfortunately, if you have monitored dozens of tiny applications with a 3S or 10S plan before, then this pricing change will affect you the most. We realized in the last months, that for many features, continuous processing is necessary at the application level and unlimited applications is therefore not scalable for us with the current pricing. You will be able to support 4 applications (1 basic license + 3 staging licenses) for 69€/month if you are fine with just the bare minimum of monitoring visibility. Or you can just stay in the old pricing if it works better for you.

What about my staging, testing, QA and development environments? I have created additional applications in Tideways for them.

Both the Basic and Standard plan include 3 staging applications for free that you can use to collect this information without having to pay extra. This allows you to create applications for development, staging and testing environments. A staging application has 60 minutes retention, but has all the same features as a basic or standard application.

How is it possible to offer a fixed rate for unlimited monitoring on unlimited hosts?

The idea of unlimited request and server plans is in our minds for a while already, but it required a greater part of the backend engineering effort last year to support this offer. Our new data structures and processing pipeline is designed in such a way that a simple and lossless aggregation per minute is possible directly when collecting data on the daemon running on your server and again when we accept the data on our backend using an aggregation proxy. This is possible with the way we use HDRHistogram and several storage optimizations as explained in this previous post on our blog. Almost no aggregation or computations are necessary for data on our backend, we just store them on disk and index them in a database. We also automatically reduce the amount of monitoring detail for the long tail of transactions with very few requests. This way it doesn’t make much of a difference in terms of processing and storage if you have just one host, or hundreds of them. It does have the downside that monitoring data is only application-centric, not on the server-level, but we think that application-centric monitoring is much better suited to understand your the experience of your users anyways!

Benjamin Benjamin 03.03.2017