Q2 Cheat sheet — SPICE at scale

All figures verified 2026-08-09 against the AWS General Reference quota table and the Quick user guide. Re-verify before quoting in a support case.

Sizing formula

Logical size (bytes) =
  (numeric cells × 12)
+ (date cells    × 12)
+ Σ (24 + UTF-8 byte length) per text cell

Computed after type transforms and calculated columns, at dataset save. Changes made in an analysis do not change SPICE size. Lat/long are numeric; all other geospatial categories are strings.

⚠️ Single dataset only. Do not sum these to estimate account capacity — read the admin SPICE capacity page.

Rule of thumb: a text column costs ≥ 24 bytes/row before its content. Dropping one 20-byte text column from 1 B rows saves ≈ 44 GB.

Quotas — the adjustable column is the one that matters

Quota	Value	Adjustable
Rows / dataset — Standard	25,000,000	❌
Size / dataset — Standard	25 GB	❌
Rows / dataset — Enterprise	2,000,000,000	❌
Size / dataset — Enterprise	2 TB	❌
`CreateIngestion` calls / 24h — Enterprise	32	❌
`CreateIngestion` calls / 24h — Standard	8	❌
Columns per file / fields per dataset	2,000	❌
Column name length	127 Unicode chars	❌
Field length (classic / new data prep)	2,047 / 65,534 Unicode chars	❌
Files per S3 manifest	1,000	❌
Visual query timeout	120 s	❌
Dataset preview timeout	45 s	❌
Calculated field expression	250,000 chars	❌
Custom actions per visual	10	❌
Email aliases per group (reports)	5,000	❌
Schedules per dataset (console)	5	—

Rows OR size — whichever you hit first. Row-level security does not reduce footprint. CreateIngestion window is floating 24h, not calendar.

aws service-quotas list-service-quotas --service-code quicksight --region us-east-1 \
  --query 'Quotas[?Adjustable==`false`].{Name:QuotaName,Value:Value}' --output table

Service code is quicksight, not quick. Not in the table ⇒ not adjustable.

Two ceilings — never confuse them

Per-dataset quota (2 B rows / 2 TB)	Account + Region capacity
❌ Not adjustable	✅ Buy more
→ redesign	→ purchase order

First-line diagnosis — run this before anything else

aws quicksight list-ingestions \
  --aws-account-id $ACCOUNT --data-set-id $DATASET --region $REGION --max-results 100 \
  --query 'Ingestions[].{When:CreatedTime,Status:IngestionStatus,Src:RequestSource,
      Type:RequestType,Secs:IngestionTimeInSeconds,In:RowInfo.RowsIngested,
      Dropped:RowInfo.RowsDropped,Err:ErrorInfo.Type}' --output table

⚠️ Alert on RowsDropped > 0, not just FAILED. COMPLETED + dropped rows = silently wrong dashboard. ⚠️ RequestType: EDIT ⇒ someone saved the dataset and triggered a reload.

Starting an ingestion

aws quicksight create-ingestion --aws-account-id $ACCOUNT --data-set-id $DATASET \
  --ingestion-id "manual-$(date +%Y%m%dT%H%M%S)" \
  --ingestion-type FULL_REFRESH --region $REGION

IngestionType: FULL_REFRESH | INCREMENTAL_REFRESH. IngestionId 1–128 chars ^[a-zA-Z0-9-_]+$. Always timestamp it — it's a PUT and ResourceExistsException is real.

HTTP	Exception	Retry?
401	`AccessDeniedException`	No
400	`InvalidParameterValueException`	No
409	`LimitExceededException`	No — terminal for the day
409	`ResourceExistsException`	New ID
404	`ResourceNotFoundException`	Check Region
429	`ThrottlingException`	Yes, backoff
500	`InternalFailureException`	Once, then case

Error triage — the 6 to know cold

`ErrorInfo.Type`	Action
`ACCOUNT_CAPACITY_LIMIT_EXCEEDED`	Buy / release capacity
`DATA_SET_SIZE_LIMIT_EXCEEDED`	Redesign — not adjustable
`SQL_SCHEMA_MISMATCH_ERROR`	Edit + save the dataset to refresh its schema
`UNRESOLVABLE_HOST` / `UNROUTABLE_HOST`	DNS / routing + SGs — different fixes
`INGESTION_SUPERSEDED`, `REFRESH_SUPPRESSED_BY_EDIT`	Not a defect — don't page
`INTERNAL_SERVICE_ERROR`	Retry once → support case with `RequestId`

Full 45-value enum: ErrorInfo

Refresh

Incremental = delete-and-replace within the window. Query window → delete SPICE rows in window → insert. Enterprise only, SQL-based sources.

⚠️ Nothing outside the window is ever updated. Late corrections and hard deletes with old dates are invisible until a full refresh. Pair every incremental schedule with a periodic full refresh.

Rule	Value
Schedules per dataset	5
Timing accuracy	within 10 minutes of scheduled time
Full refresh	Daily / Weekly / Monthly · Hourly = Enterprise only
Incremental	15 min / 30 min / Hourly / Daily / Weekly / Monthly
Hourly & sub-hourly	Exclusive — delete other schedules first
Delete incremental config	Triggers a full refresh and wipes incremental config
29th–31st monthly	Use "Last day of month"

Window on the column that changes when the row changes (updated_at), not on the event date.

Capacity

Per Region, shared by all users in the account. Bundled (not releasable) + Purchased (releasable if unused). Release order: remove/convert datasets first, then release.

Auto-purchase: console-created accounts on; API-created and pre-existing off. No auto-decrement. Off + over capacity ⇒ refreshes fail. Turning off converts current usage to purchased capacity.

⚠️ The Quick console Region selector is independent of the AWS console's. Read the Region from the URL before buying.

Support case checklist

Account ID · Region · Dataset ID · Ingestion ARN · ErrorInfo.Type verbatim · RequestId · UTC timestamps · RowInfo · what you ruled out · one answerable question.

Case type	For
Service limit increase	Adjustable quotas only
Technical	Ingestion failures, `INTERNAL_SERVICE_ERROR`, behaviour
Account and billing	Capacity charges, subscriptions

Severity	Code	First response
General guidance	`low`	24 h
System impaired	`normal`	12 h
Production system impaired	`high`	4 h
Production system down	`urgent`	1 h
Business-critical system down	`critical`	<30 m Business+ · <15 m Enterprise · 5 m Unified Ops

Stale dashboard = normal/high. Not urgent. You can raise it later on paid plans.

Direct query gotcha

"not all database drivers react to the 2-minute timeout, for example Amazon Redshift … the query runs for as long as it takes for the response to return"

Dedicated WLM queue + statement timeout. The client-side timeout does not protect the cluster.