Overview
For ultimate scalability, Deadline 8 introduces on-demand metered licensing, which will work with both local and cloud-based farms alongside existing permanent and temporary licenses. Usage is tracked per minute, and can be purchased and managed round-the-clock via Thinkbox’s e-commerce portal. Additionally, the store will host on-demand per-minute licenses for select third party content creation applications and plugins.Deadline 8 also adds a Proxy Server application that allows users to securely connect (SSL) to and interact with remote office and cloud-based render farms over public Internet without the need for a virtual private network (VPN). An updated user interface with enhanced interactivity improves the Monitor experience, while the new sandboxed Python environment facilitates additional rendering and event stability.
Highlighted Features
On-Demand Metered LicensingFor ultimate scalability, Deadline 8 introduces on-demand metered licensing. The feature will work with both local and cloud-based farms alongside existing permanent and temporary licenses. Usage is tracked per minute while the Slave application is running, and can be purchased and managed round-the-clock via Thinkbox’s e-commerce portal (coming soon). Additionally, the store will eventually host on-demand per-minute licenses for select third party content creation applications and plugins.
Deadline Proxy Server
The new Deadline Proxy Server application allows users to connect to and interact with remote office and cloud-based render farms without the need for a direct connection to the Database or Repository. The Proxy Server implements a standard REST communication protocol, allowing existing systems like Apache and Nginx to be used for SSL security and load balancing.
For example, by using Nginx and SSL security, users can connect to a cloud-based farm over public Internet without the need for a virtual private network (VPN). In addition, multi-location studios can also use the Proxy Server within their VPN to connect to Repositories in other offices without having to mount the Repository server locally. This is especially useful over high-latency network connections.
Improved Monitor Interactivity
The Monitor now remains fully interactive while loading and updating tens of thousands of jobs and thousands of slaves and data loading and updating is much quicker as well. This was accomplished by moving a lot of the list processing code to native C++ libraries.
The Monitor also remains fully interactive when processing long operations, like deleting thousands of jobs. Long running operations are queued up and shown in the new Background Operations panel, and can be cancelled or paused to allow other background operations to proceed.
Easily Launch Multiple Monitors
The Launcher and Monitor now have options to launch additional Monitors connected to other Repositories. When launching a Monitor that is connected to another Repository, it won’t change your default Repository unless you choose to, which means that other Deadline applications like the Slave won’t accidentally connect to the wrong Repository when they launch.
Asynchronous Balancer Operations
The Balancer now performs the main VM balancing operation in a separate thread. This thread can be interrupted if the interval between balancing operations has passed, or if the Balancer is being asked to shut down. This update allows the Balancer cycles to be performed at more regular intervals than before.
Python Sandboxing
Render plugins and events now run in separate processes from the main Deadline applications. This improves stability because crashes or deadlocks that occur in the render plugins and events no longer affect the Deadline applications that are executing them. For example, if the render plugin that the Slave has loaded crashes, only the sandboxed Python process will be affected, and the Slave will detect this and recover by simply starting up a new sandbox process.
When rendering, the sandboxed process’ initial environment will match the environment variables specified for the job (if there are any). In addition, the environment for render plugins is now reset between jobs, and modifications to the render environment no longer affect the Slave application’s environment. This means that environment variables set by the render plugin for one job won’t affect the next job that the Slave picks up.
Local Python API Synchronization
3rd-party Python APIs that Deadline uses are now installed and synchronized locally to the Client machines. This improves the performance of any plugins (rendering, events, cloud, etc) that rely on 3rd-party libraries because they can now be loaded off the local machine instead of from the Repository. The upgrading of Python APIs is handled by the same system that Deadline uses to perform automatic upgrades of the Client software.
Plugin and Event Configurations in Database
Rendering plugin and event plugin configurations are now stored in the Database instead of the dlinit file (similar to how Cloud plugin configurations are stored). In addition, Rendering plugin icons are now stored in the Database, and can be set via the Plugin Configuration dialog in the Monitor.
Opt-in Events
Event plugins now support a third “Opt-in” state, in addition to the existing Enabled and Disabled states. When an event is in the Opt-in state, slaves and jobs will only trigger the callbacks if they have “opted in” to the event plugin. Opt-in events can be specified for jobs in the job info file, and these events can also be configured from respective property dialogs for jobs and slaves in the Monitor.
Event plugins also support a new EventCallbacks entry in the plugin configuration that specifies which callback(s) the event triggers for.
Updated to MongoDB 3.0.6
Deadline now ships with MongoDB 3.0.6, with version 3.0.0 being the new minimum requirement for Deadline 8. Deadline utilizes MongoDB’s new collection-level locking functionality, which simplifies the data layout because data in the database no longer needs to be split up into separate databases.
If you are interested in joining the Beta, please contact
More about Thinkbox Software