Welcome to part 2 of our discussion around hybrid deployments – i.e. partially on-premises and partially in the cloud – for Media Workflow solutions. Here’s the first post, in case you missed it. We looked at both the benefits and technical trade-offs when transitioning to cloud-based solutions, and how a hybrid architecture may provide the balance you are looking for. We also looked at the multisite model in detail.
In this post we will consider a different approach to hybrid cloud architecture for Media Workflows – distributed deployments.
The distributed model
The design of a distributed Dalet Flex architecture begins with the deployment of a Dalet Flex instance either on-premises or in the cloud. When it comes to supporting cloud infrastructure providers, Dalet Flex works on Amazon Web Services, Google Cloud Platform, Microsoft Azure and Alibaba Cloud, making it one of the most open and integrated Media Workflow solutions in the market. Once the initial deployment is in place, we will leverage Flex’s micro-services architecture to implement some specific services in additional locations.
Let me give you a use case to make this clearer…
The on-premises satellite
So, you’ve decided to migrate your content supply chain a cloud infrastructure. As we mentioned in Part 1, this could be a business-driven decision, to reduce upfront investments and transition to OpEx and subscriptions costs, or a technical decision to accommodate for the increasing number of users working remotely, be it from home or other locations.
This, however, often brings new challenges, especially for production workflows, where you may be ingesting large high-resolution files on-premises, pre-existing tools deployed in your local infrastructure (QC, transcoding, editing…). Storing such files in the cloud may not be your best option: in addition to performance issues, egress costs (billed by most cloud providers when content stored in the cloud is read by external systems) can be prohibitive.
So you really need to get the best of both! Do consider the following architecture:
- The Dalet Flex Core is in the cloud. This includes databases, storage for proxy files and all the applications that power both the front-end and APIs.
- Your production storage will remain on-premises, together with limited computing power running the specific Dalet Flex micro-services that manage high-resolution files and create proxy versions sent to the cloud, so that users can search and preview content in FlexMAM.
The micro-services deployed on-premises will be fully integrated into the cloud-hosted Flex Core, and can be driven with Flex’s workflow engine just as if they were cloud-hosted too! This hybrid deployment is typically implemented to manage on-premises’ content lifecycle, together with smart archiving processes.
Alternatively, you may want to leverage cloud-based Artificial Intelligence services, like Dalet Media Cortex on the cloud-based proxy files, and trigger specific actions (e.g. archive, task allocation, transcoding….) on the corresponding on-premises high-res content.
All in all, the fact that micro-services are spread across different cloud and on-premises locations is transparent for both Dalet Flex and your end-users.
What’s the catch?
When considering a distributed deployment, you should keep in mind that solid and secure cloud connectivity for the on-premises services to connect to the cloud instance is a must.
In addition, whereas in a multisite deployment on-premises and cloud services can run independently, this is not the case in a distributed deployment. If a network outage interrupts the connectivity between the on-premises site and the cloud, local users will not be able to access FlexMAM and on-premises services will not be driven by the workflow engine as expected. To mitigate such risks, we can work with you on Disaster Recovery workflows to keep you working during a possible outage. Of course, such DR transition would be smoother in a multisite architecture (see part 1 of this blog post), which brings additional redundancy, but at a higher cost.
Another point to consider is the need to access high-resolution content from remote locations, for example when video editors need to work from home. An option is edit on the proxy, and render on the local on-premises site… but if editing on the high-res content is required, this will increase the load on the internet connection that serves your premises.
And so… to Cloud or not to Cloud?
If you are pondering over the technical design of your current, or next, Media Workflow solution, the choice between on-premises or cloud is not mutually exclusive. Hybrid solutions provide a gradual transition path to the cloud, be tailored to your business and technical requirements, and optimize your costs.
All options come with benefits and trade-offs, which we have briefly summarized in this 2-part blog post. If you’re ready to dive deeper, the Dalet engineering team can help you explore best practices and recommend an approach that meets your business priorities and technical constraints.
To find out more about how Dalet Flex can transform your media workflows, get in touch.
Over the past decade, Mathieu has built first-hand experience delivering broadcast projects across Europe and Asia, going on to leading Dalet's EMEA/APAC's Professional Services team. In his current role as VP, Product Management, Mathieu is responsible for shaping Dalet's new generation of solutions.More Articles By Mathieu