Skip to content

SYSTEM Cited by 1 source

ProGlove Insight

What it is

ProGlove Insight is ProGlove's AWS-hosted SaaS platform that connects smart-wearable barcode scanners (for frontline workers in manufacturing, logistics, retail) to digital workflows. Insight delivers real-time process visibility: scan events, productivity metrics, error rates, and ergonomics analytics back to the customer.

The system design interest is not the scanner-to-cloud protocol but the deployment architecture: Insight is the canonical production reference for account-per-tenant SaaS at scale on AWS.

Scale

At the time of the 2026-02-25 post:

  • ~6,000 AWS accounts in production (one per tenant).
  • 3-person platform team operating the whole fleet.
  • >120,000 deployed service instances across all tenant accounts.
  • ~1,000,000 Lambda functions in production.

(Source: sources/2026-02-25-aws-6000-accounts-three-people-one-platform)

Architecture in one line

Tenant → AWS account → full microservice stack (primarily serverless — Lambda, DynamoDB, managed messaging) deployed per-tenant via CodePipeline → CloudFormation StackSets from a central Infrastructure account; account creation automated via Step Functions; baseline guardrails enforced via Organizations SCPs + strict IAM; telemetry forwarded to a central third-party observability application with multi-alerts defined once and applied across tenant accounts.

Why the shape works at 3-person ops

  • Scale-to-zero service mix — idle tenant accounts cost ~0 because compute + storage are serverless.
  • Single monorepo + single pipeline — shared code versions enforced centrally; per-tenant drift is structurally prevented.
  • Account boundary = isolation + attribution + quota — every "would have been a multi-tenancy bug" class of problem simplifies to an account-level concern.
  • Platform engineering investment pays as linear, not O(n_tenants), growth — every new tenant adds marginal operational load because the platform absorbs the exponential scale. (patterns/platform-engineering-investment)

Deliberate manual steps

  • Account retirement / closure — manual scripts run regularly, deliberately not elevated to a full Step Functions workflow. Criterion: overhead of automation exceeds the benefit at retirement cadence. (patterns/automate-account-lifecycle)

Co-authorship / positioning

The 2026-02-25 post is co-authored by Julius Blank (ProGlove) and the AWS Architecture Blog. Narrative style is lessons-learned retrospective, not service-team disclosure or postmortem.

Stub page — scanner protocol, edge firmware, and non-AWS stack are out of scope for the sysdesign wiki.

Seen in

Last updated · 200 distilled / 1,178 read