Amperity Managed Production Workflows

What is “Delivery Promise”?

Delivery Promise is the commitment Amperity makes to you that, if you provide the raw resources (data drops, API availability, etc.) by the agreed upon start time, Amperity will process your data through our daily production workflow and deliver it back to you by the agreed upon delivery time. We call this time the “delivery deadline”, and it’s something we take great efforts to protect.

The two-sided nature of this promise is important - if source data is not available, it puts the entire workflow at risk and we can no longer guarantee the agreed-upon delivery time.

How does Amperity protect Delivery Promise?

Protecting delivery promise and the trust of our customers is critically important to us, and many teams across Amperity have a role in our daily operations.

  • The Delivery Operations and Support Team is responsible for reviewing, validating, and approving candidate delivery workflows to determine if they are ready for Delivery Promise. This includes reviewing workflow runtime, internal documentation, and Production Agreements to ensure that they’re correct and supportable.

  • A 24/7 rotation of on-call Support Engineers monitor, maintain, and recover delivery workflows to ensure that issues are promptly triaged, resolved and workflows recovered. This team is also responsible for customer communications during the recovery of delivery workflows.

  • In addition, the Amperity engineering team maintains their own 24/7 rotation of on-call engineers responsible for monitoring and maintaining the operations of the Amperity platform, and for responding to escalations from the on-call Support Engineer.

High-Level Overview

The following agreements detail the operations of an Amperity managed production workflow:

  1. Your account team will work with you to document in a Production Agreement the inputs, outputs, and timing of your workflow: the required and optional source files, when the workflow will begin, and by what time we will commit to delivering results. We will also document escalation contacts for us to reach out to if any required files or inputs are missing or inaccessible.

  2. Each day, you will provide source data in the agreed-upon file storage location or API-accessible system by the agreed-upon start time and using the agreed-upon file naming schemes, formats, data schema, and encryption, as detailed in the Production Agreement.

  3. We will then collect and begin processing data at the agreed upon daily start time. If any errors are encountered that require your intervention, the workflow will pause and you will be notified, either automatically or via the responding Support Engineer, via the previously documented contacts. Details of how you will be notified and how we will respond to different kinds of failures will be detailed in your Production Agreement. The workflow will then resume once we are notified that the issue has been resolved or when instructions are given to skip any missing data. Data sources may be pre-declared as “optional”, which will be skipped in the case of error, and we will use the previously-ingested data tables for those sources.

  4. If all the above criteria are satisfied, we will deliver the outputs agreed-upon with the customer at or before the agreed-upon delivery time.

If, for any reason, the prerequisites for running the daily production workflow are not satisfied, on-time delivery is not guaranteed. We will, however, continue workflows on a best-effort basis and delivery as soon as possible.

Data Processing Timelines

The length of each customer workflow is determined by the number and scale of data ingested, the types of operations performed in the Amperity Stitch, merge, and database generation process, and other factors. Your account team will work with you to determine how much time is required to process your data to determine appropriate start and delivery times. In order to be eligible to become a managed workflow, the following conditions must be met:

  1. Enough time must be allowed between the scheduled start time and the delivery deadline to allow for all of the stages of the workflow to be completed - ingest, stitching, database generation, AmpIQ insights, orchestration, etc. - while also allowing for the longest step to be retried at least once. For example, if your workflow took 6 hours, with the longest step taking 2 hours, a minimum of 8 hours must be allowed from start to finish.

  2. You must provide email contacts for us to notify when issues have been detected that need their intervention. These contacts should be available 7 days a week and ready and enabled to respond promptly when errors are encountered. If issues cannot be promptly resolved, it will put your workflow at risk of failure.

  3. Workflows that require monitoring and recovery outside business hours (Pacific time zone) will be subject to additional testing, validation, and monitoring. Your account team will discuss these details with you as appropriate.

  4. AmpIQ V1 workflows are also not eligible for monitoring and recovery outside business hours (Eastern time zone), and will be recovered the following morning.

Notification of Changes

Notification of any changes to the source data files or APIs provided by the customer must be provided one week ahead of time, or on-time delivery is not guaranteed. Examples of changes that require pre-notification include, but are not limited to:

  • Addition or removal of source data files

  • Changes to file formats (CSV to TSV, etc) or encoding

  • Changes to file name templates

  • Changes to the encryption keys used to encrypt data

  • Changes to the access credentials or permissions required to access data in customer blob stores or APIs

You should notify Amperity of changes to source data in advance. Late changes to data sources due to an urgent need or unexpected change may be possible, but we may not be able to meet delivery promise while changes are being made.

Production Support

Workflow Monitoring, Notifications, and Operations

The following guidelines will apply to the running of the production workflow:

  1. Whenever possible, you should email to alert us of source file or API delays or unavailability.

  2. When we detect required ingest files are missing, corrupt, or empty, we will notify you. In the case of optional data sources, Amperity will attempt to continue the workflow as usual.

  3. If issues arise that will prevent Amperity from meeting the daily delivery deadline, we will notify you. Depending on the process agreed upon, Amperity will either continue our workflow and deliver data when it is complete, or cancel the workflow for the day. If necessary, Amperity will also follow an agreed-upon Recovery Process to make sure that data for subsequent workflows is processed correctly.

Customer Support & Platform Operations

Amperity Customer Support

Amperity Customer Support is dedicated to providing responsive, high-quality assistance to Subscriber inquiries regarding use of the Amperity Subscription Services. Amperity support processes and tools are designed to meet the requirements of enterprise-level computing environments, and Amperity’s goal is to optimize Subscriber’s use of the Subscription Services for reliability and performance by providing effective assistance when, and if, incidents occur Details of Amperity’s Customer Support Policy, including severity levels, targeted response times, and how to contact support can be found in the Amperity Support One Pager.

Amperity Platform Monitoring & Operations

Our engineering team monitors the availability and performance of our platform 24/7 and is staffed to respond to operational issues as they occur. Most issues are discovered automatically by our system monitoring, and in most cases our engineering team will be aware of platform issues before they are discovered by you. As appropriate, Amperity Customer Support will escalate technical issues to Tier 2 and Tier 3 support within the Engineering team.