Architect track

Scale & integration architecture

Patterns for volume, resilience and loose coupling.

3 sections · 16 entries

Integration pattern picker

Match the pattern to latency, volume and who owns the source of truth.

Need	Pattern	ServiceNow tooling
Real-time create/update, low volume	Synchronous REST	Scripted REST / Table API, RESTMessageV2
Must not lose messages	Async + queue + retry	Events, async BRs, Flow with error handling, outbound queue table
Bulk nightly load	Batch import	Import Set API, data sources, transform maps, concurrent imports
Reach on-prem systems	Proxy through MID Server	MID Server, ECC queue, IntegrationHub spokes
Common SaaS (Jira, Teams, Azure…)	Spoke, not custom code	IntegrationHub spokes and Workflow Studio
CMDB from external tools	Connector + IRE	Service Graph Connectors
Stream of events from monitoring	Event ingestion	Event Management, alerts, aggregation rules

Scale patterns in code

Snippets for work that has to survive volume.

Event on a dedicated queueHidden gem

A fifth eventQueue argument routes events to a named queue so heavy work does not block the default one. Pair it with a scheduled job that processes that queue.

// Business rule (after/async)
gs.eventQueue('x_rsa.sync.requested', current, current.getUniqueValue(), '', 'x_rsa_sync_queue');

// Script Action on x_rsa.sync.requested does the outbound call,
// records attempt count and next_retry, and re-queues on 429/5xx.

Chunked scheduled job

Process a slice per run and remember where you stopped. Safe to re-run, never times out.

var BATCH = 500;
var lastId = gs.getProperty('x_rsa.cleanup.last_sys_id', '');
var gr = new GlideRecord('x_rsa_staging');
gr.addQuery('processed', false);
if (lastId) gr.addQuery('sys_id', '>', lastId);
gr.orderBy('sys_id');
gr.setLimit(BATCH);
gr.query();
var last = '';
while (gr.next()) {
  // ...process one row idempotently...
  gr.setValue('processed', true);
  gr.update();
  last = gr.getUniqueValue();
}
gs.setProperty('x_rsa.cleanup.last_sys_id', last); // '' when done = start over

Scale & resilience rules

What changes when a solution goes from pilot to enterprise volume.

✦Async by default. Anything slower than a few hundred milliseconds leaves the user transaction.
✦Idempotency keys. Correlation ID + upsert so retries and replays never duplicate.
✦Backpressure. Respect 429s and Retry-After; cap concurrency per target system.
✦Dead-letter. After N failures park the message, alert a group, allow one-click replay.
✦Archive & rotate. Archive rules and table rotation for high-volume logs and staging tables.Hidden gem
✦Transaction quotas. Know the quota rules that cancel long REST and UI transactions; design under them.Hidden gem
✦Load test. Rehearse peak volume in a sub-prod before go-live, not after the first incident.

Snippets are provided as-is under the MIT licence, without warranty. Test in a sub-production instance first. ServiceNow is a trademark of ServiceNow, Inc.; this page is independent and not affiliated with or endorsed by ServiceNow.