From February through October 2024, Appian ran a beta program for our new Autoscale feature (formerly known as EPEx). Autoscale enables Appian to scale to handle high-throughput, unattended automation processes—or, in other words, high-volume processes that don’t require human intervention. It also lets the process engine accommodate varying workloads and spikes in demand without compromising performance or requiring downtime.
With the help of eight participating organizations, the beta program achieved its goals of improved throughput and latency.
- Throughput: Beta participants saw up to 6x more process instances per minute with Autoscale (e.g., A participant processed 2,100 rows of data per minute with Autoscale compared to 350 rows of data per minute without Autoscale).
- Latency: Beta participants also processed data up to 12x faster with Autoscale (e.g., A participant processed 20,000 insurance policy purchase requests with an average processing time of 3 minutes per policy with Autoscale compared to 35 minutes without Autoscale).
Here’s what beta participants had to say:
“This beta program was 200% worth it. It has given us tremendous value, and from day one we achieved what we wanted.”
Appian Partner
“The performance improvement has been significant. Our data processing workflow will now be much easier in Appian.”
Financial Services Customer, EMEA
“Even in beta, Autoscale is already more than fast enough. This is way faster than our current tool.”
Financial Services Customer, US
“If we had Autoscale last year, we definitely would have used it instead of our current high-code implementation.”
Insurance Customer, APAC
Key takeaways and product updates
Based on the feedback from beta participants, we have focused on providing a better monitoring experience and optimizing high-throughput capabilities.
|
Monitoring improvements
- Autoscaled Process Activity tab: We added the Autoscaled Process Activity tab to give users a better view of autoscaled process data, including all child processes and errors. The data is represented as KPIs and aggregations that show the number of processes started and completed as well as the average process completion time.
- Monitoring and process history: We added and improved monitoring capabilities to make it easier to understand how an autoscaled process runs. These capabilities improve the debugging experience by visualizing the process flow and showing changes in process variable values.
|
High-throughput improvement
- Smart services and nodes: Beta participants helped influence which smart services and nodes are supported for the initial release of Autoscale, including the start process and call integration smart services.
- Ease of using Autoscale with process models: Designers will have access to a wide variety of process functionality in the new engine, enabling many process models to leverage Autoscale as easily as checking a box.
- Infrastructure and roadmap: Beta participants helped with infrastructural hardening while also contributing to roadmap prioritization around new methods for process kickoff (e.g., Kafka).
|
If you are interested in using our new Autoscale feature, please contact your account team or complete this form.