When it comes to feature flags, there are several packages available for start-ups on a tight budget. Here's a comparison of cross-platform packages which all offer flag management, segment targeting and percentage-based targeting.
It's worth pointing out that these services could be used for remote configuration, but they don't really compare to Firebase Remote Config. The latter is better suited to modifying content remotely. While none of these packages have out-the-box A/B testing analytics, APIs and integrations are available.
Flag evaluations are implemented differently on each platform:
- Cloudbees Rollout and ConfigCat have opted to implement all evaluations on the client side, their servers only store flags and rules.
- LaunchDarkly and FeatureFlow have two types of SDK's. Server-side SDK's are comparable to those of ConfigCat and Rollout. In contrast, client-side SDK's allow flag evaluations to take place on the server-side.
- Bullet train does all feature flag evaluations on the server-side.
Arguably, server-side evaluations are better suited for mobile apps and web front ends, whereas client-side evaluations are fast and secure.
A differentiating factor is the functionality to schedule releases through the console, LaunchDarkly and FeatureFlow have incorporated this into their front end. Another front-end feature of interest is user segmentation management, which is available with LaunchDarkly, Rollout, and Bullet train subscriptions.
Let's dive into the available options below.
LaunchDarkly
Well established, boasting many SDKs, integrations and a functional web app.
The "Starter" package can be purchased for 75$ per month and allows for unlimited projects, environments, feature flags and rules. Reach, however, is capped at 1000 client-side monthly active users (MAUs); Server-side MAUs are uncapped. In addition, teams are limited to one developer / product owner.
LaunchDarkly probably supports more languages than any competitor product on the market. They also support JSON flag types. As mentioned, both segments and scheduled releases can be managed through their web app as an additional convenience.
SDKs: Android (Java and Kotlin), C/C++, Electron, Erlang, Go, Haskell, iOS (Objective and Swift), Java, JavaScript, Node.js, PHP, Python, React, React Native, Ruby .Net, Roku, Xamarin
Integrations: Azure DevOps (formerly VSTS), Bitbucket Pipelines, Datadog, Dynatrace, Jira, Microsoft Teams, New Relic, OAuth, Slack, Splunk, Terraform, Visual Studio Code, Webhooks
ConfigCat
Ideal for large teams with few applications and variable MAUs.
ConfigCat's start-up package, "Pro" goes for 49$ per month if you sign a 12-month contract. There is no limit on the number of users who can be organised into 2 permission groups.
There is also no limit on the number of MAUs, although, the service is capped at 25 million requests per month - that's about 10 requests per second. As per the SLA, they guarantee 99.8% uptime. Support is also available on slack and live chat over and above email.
The package, however, only allows for 100 feature flags, 3 products and 3 environments. Each flag can have a maximum of 8 segment rules and 8 percentage-based rules.
SDKs: Android (Kotlin), Angular, Go, iOS (Swift), Java, JavaScript, .NET Node.Js, PHP, Python, React, Ruby
Integrations: CircleCI, DataDog, GitHub Action, Microsoft Teams, Slack, Trello Power-Up, Webhooks (3 per environment), Zapier
Cloudbees Rollout
Perfect for small scale (streaming) apps and A/B testing.
With an annual contract, the "Solo" package will set you back by 60$ per month. This package includes unlimited feature flags, projects, environments and rules. An obvious benefit is the ease with which you can dive into A/B testing analytics using Google Analytics.
The service works with streaming applications and is compatible with Android TV, Fire TV, Roku, Chromecast, Apple TV and Tizen.
MAUs are, however, capped at 2000 and as the package name suggests, only one user has access to the platform.
SDKs: Android (Java), Go, iOS (Swift and Objective-C), Java, Javascript, .NET, Node.js, PHP, Python, React Native, Ruby, Typescript
Integrations: DataDog, Google Analytics, Jira, Mixpanel, New Relic, Prometheus, Segment
Bullet Train
Good for low traffic applications with potentially high MAUs.
The "Scale-Up" package can be purchased for 99$ per month. It includes unlimited flags, projects, environments and rules. The service is capped at 2 million requests per month (1 request every 1.25 seconds) and a maximum of 10 team members have access to the platform. It comes with priority email support.
There is also a more affordable "Start up" package for just 29$ per month. This package limits the number of requests to 250 thousand (1 request every 10 seconds) and 3 users. The priority email is also excluded.
As the project is open source you could opt to host it free of charge.
Client Libraries: Android (Java), iOS (Swift)), Java, Javascript, .NET, NodeJS, Python, React, React Native, Ruby
Integrations: Webhooks
FeatureFlow
Functionality without the bells and whistles.
Their "Medium" plan retails for $80 per month. On this plan, you receive unlimited feature flags, projects, environments and rules. It also comes with 'Basic' support which guarantees an uptime of 99.8%.
This plan allows teams of up to five to use the service. Unlike competitor products, audit logs are not included. In addition, MAUs are limited to 10 000.
FeatureFlow offers a lightweight "Playground" option for 1 user. This can be used for 1 project, with maximum 2 environments and 10 feature flags. MAUs are capped at 1000.
Integration and language support are rather limited.
SDKs: AngularJS, Go, Java, JavaScript, .NET, Node.Js, React, Ruby
Integrations: Slack
The table below summarises the pros and cons of each package.
In summary, LaunchDarkly's 'Starter Package' supports the most SDK's and their web interface is slightly more functional. ConfigCat's "Pro" package allows large teams to work together. Rollout's Solo package is the most convenient for A/B testing. Bullet Train's "Scale-Up" package is suitable for low traffic applications. FeatureFlow's 'Medium' package is ideal if you don't need added features such as audit logs and two factor authentication.
| LaunchDarkly: “Starter” | ConfigCat: “Pro” | Rollout: “Solo” | Bullet Train: “Scale-Up” | FeatureFlow: “Medium” | |
|---|---|---|---|---|---|
| Cost ($) | 75 | 49 (annual contract) | 60 (annual contract) | 99 | 80 |
| Number of users | 1 | Unlimited (2 permission groups) | 1 | 10 | 5 |
| MAUs | 1000 client side, unlimited server side | Unlimited | 2 000 | Unlimited | 10 000 |
| Number of projects | Unlimited | 3 | Unlimited | Unlimited | Unlimited |
| Number of environments | Unlimited | 3 | Unlimited | Unlimited | Unlimited |
| Number of feature flags | Unlimited | 100 | Unlimited | Unlimited | Unlimited |
| Number of rules per feature flag | Unlimited | 16 per flag | Unlimited | Unlimited | Unlimited |
| Requests per month | Unlimited | 25 million | Unlimited | 2 million | Unlimited |
| Impressions per month | Unlimited | Unlimited | 500 000 | Unlimited | Unlimited |
| Security | Two Factor Authentication | Two Factor Authentication and SSO | Two Factor Authentication | ||
| Audit logs | Yes | Yes | Yes | Yes | No |
| Feature flag evaluations | Server side and client side | Client side | Client Side | Server Side | Server side and client side |
| Guaranteed Uptime as per SLA | 99.8% | 99.8% |
Top comments (0)