Buy or build? The honest answer.
You could build SAP document routing yourself — an ArchiveLink content server for ECC, a CMIS client for BTP DMS, an OData webhook receiver for S/4HANA Cloud, on BTP, your own VM, or with ABAP. People do. Here’s what they discover in months 3, 6, and 18 — and where FileRelay actually changes the calculus.
What “just build it” actually means
Each protocol — ArchiveLink, CMIS, OData — looks small on the surface. Production-grade scope across all three is much bigger.
Three protocols, three specs
ArchiveLink BC-HCS 4.5 (12 actions, signed URLs, certificate management, putCert/serverInfo), CMIS 1.1 Browser binding (change-log walk + content fetch + auth), OData attachment APIs (CloudEvent verification, OAuth refresh, $value paging). Build one well, you still owe two.
Per-destination integration
OAuth flows, token refresh, retry semantics, error mapping, request shapes per SDK. Build once for S3; then SharePoint; then Google Drive; then the next request.
Operational guarantees
At-least-once delivery, dead-letter queue, retry policy, idempotency, observability, replay tooling. The boring stuff that decides whether the system makes it past pilot.
Security & compliance
mTLS or pinned certs to SAP, per-customer encryption of credentials, audit log of every routing decision, retention policy, data subject access response.
SAP edge cases
Certificate rotation, late vs early archiving, barcode flows you might need later, print-list and spool archiving, FI/MM document classes with their idiosyncrasies.
Ongoing maintenance forever
SharePoint Graph API changes. AWS deprecates SigV2. OAuth scope changes. New SAP support pack. Knowledge concentrates in one engineer who eventually leaves.
Build vs. SAP BTP DMS vs. FileRelay
Apples-to-apples on the three real options. FileRelay can sit alongside BTP DMS — not instead of it.
| Build on BTP / ABAP | SAP BTP DMS | FileRelay | |
|---|---|---|---|
| Time to first document routed | 8–12 weeks | Days (SAP-managed) | < 1 day |
| Initial engineering effort | 200–500+ hours | BTP subscription + config | 1 hour Basis setup |
| Destinations supported | 1 (typically) | 1 (BTP DMS itself) | 15+ out of the box |
| Multi-destination routing (fan-out) | |||
| You choose the storage | BTP-locked | ||
| Multi-SAP-system support | Custom per system | One DMS per subscription | Workspace per system |
| Routes to BTP DMS itself | N/A | ||
| Self-hosted / on-prem | |||
| OAuth refresh / API drift | SAP-managed | ||
| Retry, dead-letter, replay | Build it | SAP-managed | Built-in |
| Audit trail (per document) | Build it | SAP-managed | Built-in, exportable |
| SAP ICC BC-HCS 4.5 cert | Your responsibility | Certified | On our roadmap |
| Pricing model | Your eng cost | Per-document, per-storage | Annual licence per deployment |
| Lock-in risk | Knowledge concentration | BTP ecosystem | Config change to swap |
SAP BTP DMS is a perfectly fine destination for many SAP customers — FileRelay can deliver to it alongside any other destination. The point is choice, not replacement.
When building still makes sense
We’d rather lose a deal than win one we’re not the right fit for. These are the honest cases where build wins.
Truly bespoke transformation
You need document-level business-logic transformation (not just routing) that nothing off-the-shelf models. Build it; we’re a poor fit.
Existing internal product
You already have a maintained internal content server with team competence and a roadmap. Don’t fix what works.
Strategic core competence
Document routing IS your product (e.g. you’re an SAP-vertical SaaS). You should own that core, not license it from us.
If your situation matches any of these, we’ll say so on the call.
The build costs nobody puts in the BTP proposal
These are real, recurring, and usually invisible until month 6.
Schedule slippage
Day-one estimate is 6 weeks. ArchiveLink protocol quirks, certificate negotiation with SAP Basis, OAuth callbacks for SharePoint — you ship in 16.
ICC certification
If procurement requires a certified content server (some do), you’ll apply to SAP ICC, run BC-HCS 4.5 tests against a Minisap, and budget 12–18 months. Then maintain it.
Bus factor of one
The engineer who built it becomes load-bearing. When they leave, the next person needs 3 months to understand ArchiveLink wire-level semantics.
Destination drift
SharePoint, S3, Azure Blob each update breaking-change-style every 12–24 months. Each one needs investigation, fix, regression test, redeploy.
Audit when it matters
A GoBD-style audit asks: who routed which document where, on what date, with what outcome? You build the answer, retroactively, under time pressure.
Scope creep
“Just to S3.” Six months later: also SharePoint. A year later: Slack notifications and email copies. The simple build becomes a small product.
Talk through your scenario
30 minutes with a Basis-friendly engineer. We'll walk through your destinations, scale, and constraints — honest answer either way.