As we continue to develop Tator, our web platform for video annotation and analysis, we have identified two distinct user groups. The first group consists of Open Source Software (OSS) Users who are typically small teams or individual researchers looking to use Tator for a single project or field work. They prefer to install Tator on a single on-premise machine, without relying on cloud services. They have limited data, typically less than 10TB, and require access for a small team of 1-10 users. On the other hand, Enterprise Users are medium to large organizations that require high availability and data durability, with scalability and security being their top concerns. They are interested in using the cloud and require access to large amounts of data, often in the range of tens or hundreds of TB.
As developers, we recognize the importance of serving both types of users, and have endeavored to do so with a single open source codebase. However, we have found that our current codebase is not optimal for either group. For OSS Users, Tator is difficult to install and configure due to its reliance on Kubernetes. Despite the development of our install script and support for microk8s, OSS Users often experience issues related to container networking, particularly with DNS, firewalls, and proxies. Meanwhile, for Enterprise Users, the value proposition of a Tator Enterprise Subscription is unclear since Tator is entirely open source.
To address these issues, our repository will be split into two separate repositories, Tator OSS and Tator Enterprise, starting with release 1.1.0, which is our next big milestone release.
Simplifying Tator OSS
Tator OSS will focus on the needs of OSS Users, providing a solution that is easy to install and configure. To achieve this, we will switch from running Tator on Kubernetes to running it with Docker Compose. This will make Tator much more portable and easier to deploy on different platforms like Docker Desktop. With this change, Tator will no longer require a "fresh" install of Ubuntu or a large number of dependencies, and it will be available via http://localhost. Even better, Tator OSS can still be used for collaborative annotation by configuring a reverse proxy like NGINX or Traefik directly on the host node.
On the other hand, Tator Enterprise will cater to the needs of medium to large organizations who require high availability and data durability, scalability, and security. By separating the codebase, Tator Enterprise's value proposition becomes clearer to users who require the added benefits of the commercial license.
Despite the split, Tator OSS will retain all the features that OSS Users have come to expect from our platform. This includes a range of annotation tools, AI/ML integrations, and project management features. To see a detailed feature matrix and compare Tator OSS and Tator Enterprise, refer to the table below.
Introducing Tator Enterprise
As we approach the Tator 1.1.0 milestone, we are excited to announce the addition of Tator Enterprise, the proprietary counterpart to Tator OSS. Tator Enterprise will continue to run on Kubernetes, and our Tator Enterprise Subscription model allows us to offer Enterprise Users a suite of new security and scalability features, in addition to the configurability that they are already familiar with. We are confident that this new approach will make it easier for Enterprise Users to understand the value of Tator.
So, what can you expect from Tator Enterprise that Tator OSS does not offer? Let's dive into the feature matrix below:
| Feature | Tator OSS | Tator Enterprise |
|---|---|---|
| Video hosting | ✔ | ✔ |
| Video streaming | ✔ | ✔ |
| Frame-accurate playback | ✔ | ✔ |
| Video and image annotation | ✔ | ✔ |
| Analytics views | ✔ | ✔ |
| Configurable metadata | ✔ | ✔ |
| REST API | ✔ | ✔ |
| Python and JavaScript clients | ✔ | ✔ |
| Algorithm registration | ✔ | ✔ |
| Custom dashboards | ✔ | ✔ |
| Automatic database maintenance | ✔ | ✔ |
| Automated emails | ✔ | ✔ |
| Media archive and restore | ✔ | ✔ |
| System announcements | ✔ | ✔ |
| Runs on Docker Compose | ✔ | |
| Serial transcodes | ✔ | |
| Session-based front-end authentication | ✔ | |
| Runs on Kubernetes | ✔ | |
| Parallel transcodes | ✔ | |
| OpenID Connect front-end authentication | ✔ | |
| Aggregated logging and metrics | ✔ | |
| Third-party authentication | ✔ | |
| Metrics-based autoscaling | ✔ | |
| Domain aliases and redirects | ✔ | |
| Backup buckets | ✔ | |
| Cross-origin resource sharing | ✔ | |
| Automated TLS certificate management | ✔ |
Tator Enterprise Subscriptions will also still include phone, email, and Slack support, as well as a commercial license that allows Enterprise Users to modify Tator for their needs.
Wrapping up
In conclusion, we hope that the new repository structure for Tator, with Tator OSS and Tator Enterprise, will better serve our diverse user base. By dividing the codebase, we can tailor Tator OSS to be more portable, easier to use, and require fewer dependencies, while still retaining all the features that OSS Users value. Meanwhile, Tator Enterprise will offer the configurability, security enhancements, and scalability features that Enterprise Users require, all under a commercial license via a Tator Enterprise Subscription. We appreciate the feedback and support of our user community, and we look forward to continuing to improve Tator for all our users.