Deep navy editorial cover — two frosted-glass server towers on a circuit plinth, one missing a layer marked in amber, under a tick-marked measuring beam; the AWS mark floats upper right and the left half stays open for the title.

AWS China vs Global: Partition Differences Measured Across 66 Services

We measured AWS China region differences with read-only APIs across Beijing, Ningxia, and us-east-1 — versions, features, quotas, and unit prices for 66 services. Most of the core is in sync; rate limits are not.

Most writing on AWS China region differences is a catalog check: a service is listed or it is not. That method misses where migrations actually fail — on load-test day when nothing launches, or in month three when the event bus throttles at 4% of the traffic it was sized for. Catalog presence says nothing about versions, sub-features, rate limits, or price.

The symmetric measurement at a glance: one read-only probe splits to the global and China partitions — the us-east-1 globe and the Beijing–Ningxia pillars — feeds a four-dimension analyzer for versions, features, quotas, and cost, and lands on three verdicts: core in sync, rate limits bind on day one, cost splits by region.

This article reports a different measurement. Two brand-new AWS accounts — one in the China partition (Beijing cn-north-1, Ningxia cn-northwest-1), one in the global partition (us-east-1) — were measured against each other across 66 services and four dimensions. Because both accounts were empty on day one, every number here is a cold-start baseline: the defaults a new product team inherits at AWS China, with no account history and no negotiated limits.

The four dimensions, per service, always the same:

DimensionThe questionWhy a product team cares
VersionsAre engine / runtime / cluster versions in sync?Migration scripts and feature assumptions break on version skew
FeaturesIs the service deployed, and are sub-features present?You discover mid-build that you have to roll your own
QuotasDo default limits match global?The limit you hit is usually a quota, not a missing service
CostWhat does the same unit cost?Budget math done with us-east-1 numbers is wrong twice

Two scope notes. First, this is not a catalog comparison; accounts, operators, and catalog gaps are covered in our AWS China orientation guide. Second, AI services are out of scope for the per-service matrix: in a separate measurement campaign, AWS China Bedrock availability resolved to a simple fact — no endpoint for Bedrock or any managed AI API in either China region, leaving SageMaker self-build, with a 2021-generation A10G accelerator ceiling, as the only path.

Official AWS sources, for cross-checking anything below:

These are the authoritative references; this article’s numbers are independent measurements taken at a point in time.

How we tested

The method — we call it a four-dimension partition benchmark — is symmetric measurement: every number below was produced by calling the same read-only AWS API, with the same commands, against both partitions on the same days, so any difference in these tables can only come from the platform, never from the measurement. The principle is falsifiability: each table names the API behind it and every open question we could not resolve is published rather than smoothed over, so your team can re-run the commands against your own accounts and check us.

  • When: October 11–12, 2026, in six batches, all commands re-runnable.
  • Accounts: One fresh China-partition account, one fresh global account. Account identifiers are excluded from this article. “Fresh” matters: applied quotas carry account history, so every quota number below is the default layer — the factory setting a new team inherits.
  • Method: Read-only API calls only — describe / list / get. Zero resources created, zero spend. Service-quotas values read from both API layers (see field note below). Engine versions from describe-*-engine-versions. Endpoint existence verified per service by calling a minimal read API and classifying the response; “not deployed” below always means the endpoint does not exist in public DNS or rejects the call as unsupported — never a guess.
  • Pricing: Read from the Price List API (EC2 per instance type; managed services pulled as full regional catalogs and matched client-side). One structural finding: RDS and Aurora compute SKUs are absent from the Pricing API for the China partition — for those, and only those, you must price from the official AWS China price list; the blind-spot table in the Cost insight is the complete list of verified gaps. Everything else in this article is API-sourced.
  • All prices are shown in USD; China list prices were billed in CNY and converted at ≈7.1 CNY/USD.

Field note: read quotas in two layers, or you will misread them

Our first pass concluded that the Lambda concurrency quota “does not exist” in China: GetServiceQuota returned NoSuchResource while the console showed 1,000. The resolution: applied-value and default-value are two different APIs, and a quota that was never initialized is invisible to the applied layer — GetAWSDefaultServiceQuota returned 1,000 cleanly. This is not China-specific; any fresh account behaves this way. Every quota claim in this article survived the two-layer read.

Key numbers at a glance

For readers with thirty seconds — and for anyone quoting this benchmark — five numbers carry most of the weight. Each links to the section that measures it.

NumberWhat it measures
25×EventBridge and SNS default throughput vs global — 750 vs 18,750 TPS invocations, 120 vs 3,000 TPS topic operations (Quotas insight)
10×KMS symmetric crypto, 10k vs 100k TPS, and Athena concurrent queries, 20 vs 200 (Quotas insight)
+48% / −1%The same m5.xlarge on-demand hour in Beijing and Ningxia, against us-east-1’s $0.192 (Cost insight)
+153%ALB LCU-hour in Beijing; Ningxia prices the identical unit at parity (Cost insight)
66 + 10Service surfaces measured in full plus surfaces absent in both regions — the closed ledger behind every claim above (The matrix)

The matrix: 66 services, four dimensions, three regions

Each service gets the same four-row table. The Gap column is the verdict; the three region columns are the evidence. “Parity” means all three regions returned the same answer. Ten further service surfaces have no endpoint in either China region; they are consolidated in the closing table of this section, so the ledger closes at 66 measured plus 10 absent. One convention on Cost rows: a number is API-measured on both sides, three dashes mean the unit was simply not captured by this benchmark, and the only verified API absence — RDS and Aurora compute — points to the blind-spot table in the Cost insight.

Four-dimension gap severity matrix for AWS China versus global. Versions and features show minor to moderate gaps in both regions, quotas are severe, Beijing cost is extreme, and Ningxia cost sits below global.

Compute and runtime

EC2 — one-third the catalog, no 7th generation, three GPU chips

Dimensionus-east-1BeijingNingxiaGap
VersionsCurrent generations (m7i / c7i / r7i / p6 present)No 7th-gen familiesSame as Beijing7th-generation instance families absent in China
Features1,375 types / 146 accelerator types432 / 33461 / 40~1/3 breadth; GPU: 13 chips vs 3 (T4, A10G, Inferentia); within-family sizes never shrink — what’s missing is whole families
QuotasCatalog 1,775 items; vCPU default 51,614 items (1,526 not adjustable)1,613 itemsDefaults match; 94% of Beijing catalog items are not adjustable
Costm5.xlarge $0.192/hr$0.285 (+48%)$0.191 (−1%)Beijing +23–49%; Ningxia at or below us-east-1, GPU −6%

EKS — service healthy; quota self-service not integrated

Dimensionus-east-1BeijingNingxiaGap
Versions1.32–1.37 (7 versions)Same, latest 1.37SameFully in sync, Auto Mode available
FeaturesFull cluster + addon APIsWorks; list-addons returns NotFoundWorksCore is there; addon management surface has a gap
Quotas12 self-service quota itemsNot integrated into Service Quotas (NoSuchResource)SameThe one genuine partition-level quota integration gap — the service itself is healthy
CostControl plane $0.10/hr; Fargate $0.0405/vCPU-hrControl plane $0.097; Fargate $0.051 (+26%)SameControl plane at parity; Fargate +26%

ECS / Fargate — API parity; the difference is quota management

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
Featuresdescribe/list all availableWorksWorksParity
Quotas62 items (default layer)62 items, identical values62 itemsDefault layer identical — measured, not assumed
CostFargate vCPU/GB billing——Priced in the compute bill (EKS Fargate SKU)

Lambda — durable-execution rates halved

Dimensionus-east-1BeijingNingxiaGap
VersionsRuntimes not measurable read-only (no standalone API)n/an/aFlagged honestly, not guessed
FeaturesWorksWorksWorksParity
Quotas75 items33 items; durable-execution rates halved (6,000→3,000 TPS family)33 itemsItem count 44%; the newest feature ships with half the throttle
Cost$0.0000167/GB-s$0.00002/GB-sSameCompute price at parity across all three regions

Auto Scaling — full parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas9 items, all identical99Identical values
CostBilled with EC2SameSame—

Batch — full parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas15 items (all not adjustable)15, identical15Identical
CostBilled via Fargate/EC2SameSame—

Elastic Beanstalk — works everywhere; quota plane absent in China

Dimensionus-east-1BeijingNingxiaGap
VersionsPlatform branches API availableSameSameDepth comparison deferred (API granularity is coarse)
FeaturesWorksWorksWorksParity
Quotas5 itemsNot integratedNot integratedQuota self-service absent
CostBilled via EC2/ELBSameSame—

Messaging and streaming

SQS — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values measured
CostPer-request———

SNS — transaction rates at one twenty-fifth of global

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasTopic ops at 3,000 TPS120 TPS across the board120 TPS25× throttle — fan-out-heavy designs must re-rate
CostPer-request———

EventBridge — invocation and PutEvents rates at one twenty-fifth

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas21 items; invocations 18,750 TPS / PutEvents 10,000 TPS8 items; 750 / 400 TPSSame25× throttle on the event backbone — event-driven capacity plans must be re-derived, not ported
CostPer-event———

Step Functions — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer transition———

MSK (Kafka) — 36 versions, fully in sync

Dimensionus-east-1BeijingNingxiaGap
Versions36 versions (4.2.x KRaft)36, same36, sameFully in sync
FeaturesWorksWorksWorksParity
Quotas7 items7, identical7Identical
CostPer broker———

Amazon MQ — the dataset’s only major-version gap

Dimensionus-east-1BeijingNingxiaGap
VersionsRabbitMQ 4.3 / ActiveMQ 5.19RabbitMQ 4.2 / ActiveMQ 5.18SameBoth brokers one major version behind
FeaturesWorksWorksWorksParity
Quotas17 items17 (15 not adjustable)17No differing values
CostPer broker + instance———

Kinesis (Streams + Firehose) — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesStreams/Firehose APIs workWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer shard/PUT———

Application platform

AppSync — API parity; quota plane absent

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIn catalogNot integratedNot integratedQuota self-service absent
CostPer request———

API Gateway — REST works; HTTP API does not exist

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesREST + HTTP APIsREST works; HTTP API (v2) endpoint not deployedSameArchitectures must use REST v1 — cheaper-to-run HTTP APIs are not an option in China
Quotas51 itemsNot integratedNot integratedQuota self-service absent
CostPer request———

AppConfig — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasNot listedNot listedNot listed—
CostPer configuration update———

Cloud Map — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer endpoint———

Systems Manager (incl. Parameter Store) — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIn ssm catalogIntegratedIntegratedNo differing values
CostStandard parameters freeSameSame—

Data layer

RDS MySQL — versions in sync; Beijing and Ningxia quotas diverge

Dimensionus-east-1BeijingNingxiaGap
Versions22 versions, latest 8.4.1122, same22, sameFully in sync
FeaturesWorksWorksWorksParity
QuotasDB instances 40; parameter groups 50DB instances 20; parameter groups 2040; 50Beijing 20 vs Ningxia 40 — an intra-China asymmetry, not a partition one
Costdb.m5.large / db.r5.xlarge baselines capturedNot in Pricing API — blind-spot listSameOfficial China price list

RDS PostgreSQL — latest in sync; 11 fewer historical versions

Dimensionus-east-1BeijingNingxiaGap
Versions56 versions, latest 18.645, latest in sync45Latest matches; fewer rollback targets
FeaturesWorksWorksWorksParity
QuotasAs RDS family20/2040/50Same Beijing asymmetry
CostBaselines capturedNot in Pricing API — blind-spot listSameOfficial China price list

RDS MariaDB — fully in sync

Dimensionus-east-1BeijingNingxiaGap
Versions27 versions, 12.3.327, same27Fully in sync
FeaturesWorksWorksWorksParity
QuotasFamily valuesFamily asymmetryFamily valuesAs above
Cost—Not in Pricing API — blind-spot listSameOfficial China price list

Aurora MySQL — latest in sync

Dimensionus-east-1BeijingNingxiaGap
Versions35 versions33, latest in sync33Latest matches
FeaturesWorksWorksWorksParity
QuotasFamily valuesFamily asymmetryFamily valuesAs above
CostAurora baselines capturedNot in Pricing API — blind-spot listSameOfficial China price list

Aurora PostgreSQL — two minor versions behind

Dimensionus-east-1BeijingNingxiaGap
Versions46 versions, 18.640, 18.440, 18.42 minor versions behind — the only Aurora line that lags
FeaturesWorksWorksWorksParity
QuotasFamily valuesFamily asymmetryFamily valuesAs above
CostAurora baselines capturedNot in Pricing API — blind-spot listSameOfficial China price list

DynamoDB — full parity, measured twice

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity; global tables do not replicate across partitions
Quotas13 items; account 80k / table 40k CU13, identical values13Fully identical
CostPer WCU/RCU———

DocumentDB — engines in sync; instance classes trimmed

Dimensionus-east-1BeijingNingxiaGap
Versions8 versions (8.0.2)8, same8Fully in sync
Features39 instance classes33 (r4-generation missing, r5 intact)2785% / 69% breadth
Quotas10 items10, identical10Identical
CostPer instance———

ElastiCache — engines in sync (including Valkey); Ningxia trims history

Dimensionus-east-1BeijingNingxiaGap
Versionsredis 6 (7.1) / valkey 6 (9.1) / memcached 11Samememcached 7Latest all in sync; Ningxia drops 4 historical memcached versions
Features236 purchasable SKUs167 (~70%)167SKU breadth 70%
Quotas12 items12, identical12Identical
Costcache.r5.large $0.216/hr$0.235 (+9%)$0.213 (−1%)Near parity; Ningxia effectively at us-east-1

MemoryDB — fully in sync

Dimensionus-east-1BeijingNingxiaGap
Versionsredis 3 (7.1) / valkey 2 (7.3)SameSameFully in sync
FeaturesWorksWorksWorksParity
Quotas8 items8, identical8Identical
CostPer node———

Neptune — the only engine ahead of global

Dimensionus-east-1BeijingNingxiaGap
Versions28 versions, latest 1.4.8.030, latest 1.4.8.130, 1.4.8.1Both China regions ahead of us-east-1 — the dataset’s only inversion
FeaturesWorksWorksWorksParity
Quotas13 items13, identical13Identical
CostPer instance———

OpenSearch — engines in sync; the serverless variant is absent

Dimensionus-east-1BeijingNingxiaGap
Versions36 versions3636Fully in sync
FeaturesIncludes OpenSearch ServerlessServerless endpoint not deployedSameManaged-serverless search is not an option in China
Quotas5 items (es)5, identical5Identical
CostPer instance———

Redshift — works; version API too coarse to compare fairly

Dimensionus-east-1BeijingNingxiaGap
Versionscluster-versions API returns majors onlySameSameGranularity flagged rather than forced
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer node———

DAX — works; a quota-catalog oddity worth recording

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIn catalogNo catalog entryIn catalogCatalog asymmetry — availability is unaffected
CostPer node———

Storage and backup

S3 (incl. S3 Tables) — parity, except Beijing’s object-size ceiling

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorks; S3 Tables availableWorks; Tables API reachable (functional check pending)WorksParity
QuotasObject size quota 48.83 TB; replication destinations 28Object size quota 5 TB; replication destinations 248.83 TB; 2Object-size figures are an API-convention discrepancy, not a confirmed gap — 5 TB matches the documented global single-object limit, 48.83 TB matches the 10,000-part × 5 GiB multipart ceiling; console review pending. Replication fan-out capped at 2
CostStandard $0.023/GB-Mo$0.027 (+19%)$0.025 (+8%)Modest premium

EBS — parity, and gp3 is cheaper in Ningxia

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity; per-volume concurrent snapshots 5 (not adjustable) in all three
QuotasUnder EC2 catalogSameSameSmall numbers identical
Costgp3 $0.08/GB-Mo$0.084 (+5%)$0.075 (−6%)Full volume-type ladder extracted

EFS — works; quota plane absent in China

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas23 itemsNot integratedNot integratedQuota self-service absent
CostPer GB———

FSx — works; OpenZFS capacity quota quartered

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas30 items; OpenZFS SSD 262 TB25 items; 65 TB25OpenZFS capacity at 1/4
CostPer GB/throughput———

AWS Backup — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas12 items9, identical values93 fewer items; values match
CostPer backed-up volume———

Network and edge

VPC / ELB — Beijing LCU premium on the ingress line

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesALB/NLB workWorksWorksParity
Quotas31 items27 (5 not adjustable)27No material differences
CostALB $0.0225/hr + $0.004/LCU-hrALB $0.022 (≈parity) + LCU $0.01 (+153%)ALB $0.014 (−37%) + LCU ≈parityBeijing’s LCU premium is the single biggest ingress cost gap; Ningxia wins both lines

Transit Gateway — works; attachment premium 41%

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorks (API verified)WorksParity
QuotasUnder VPC catalogNo separate itemsNo separate itemsFlagged for review
CostAttachment $0.05/hr + data $0.02/GBAttachment $0.07 (+41%); data ≈paritySameCross-region mesh costs more per attachment
Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
Features633 endpoint services (incl. third-party SaaS)180181~29% catalog breadth — private connectivity to SaaS partners is where the catalog thins out
QuotasUnder VPC catalogSameSameNot separately probed
CostInterface endpoint $0.02/hr + $0.01/GB$0.016 + $0.00986$0.012 + $0.00986Prices slightly below us-east-1

CloudFront — distributions work; no edge compute; Ningxia end of support 2027-05-31

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesFunctions / KV stores / realtime logs availableAll three: “not supported in this region”Same, plus published end of support: 2027-05-31Lambda@Edge unavailable by dependency; Ningxia carries a retirement date
Quotas98 items98 (43 not adjustable)No catalog entryItems match; catalog asymmetry recorded
CostPer traffic———

Route 53 — zones and health checks in; traffic policy and registration out

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesFull incl. traffic policies + domain registrationZones + health checks; traffic policies invalid; no domain registrationSameThird-party registrar + ICP filing is the path
Quotas500 hosted zones (catalog entry Beijing only)IntegratedNo catalog entryCatalog asymmetry
CostPer zone/query———

Direct Connect — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas16 items (all not adjustable)16, identical16Identical
CostPer port/transfer———

Identity and security

IAM — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas23 items23 (catalog entry Beijing only)No catalog entryValues identical
CostFreeFreeFree—

Organizations — fully usable inside China; no bridge to your global org

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorks (SCPs verified in the earlier campaign)WorksBuild parallel, don’t extend: the two orgs are disjoint
Quotas20 items0 items listed (integrated but empty on a fresh account)No catalog entryEmpty catalog ≠ unavailable
CostFreeFreeFree—

IAM Identity Center — parity within the partition

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorks (instances + permission sets verified)WorksParity
QuotasNot listedNot listedNot listed—
CostPer userSameSame—

KMS — request rates at one tenth

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasSymmetric crypto 100k TPS; RSA keypair gen 20/4/110k TPS; 1/0.5/0.1Same10× on symmetric, 20× on RSA generation — envelope-encryption-heavy designs need a caching layer
CostPer key/request———

Secrets Manager — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer secret/version———

ACM — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasNot listedNot listedNot listed—
CostPublic certs freeSameSame—

WAF (v2) — service parity; quota plane absent

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIn catalogNot integratedNot integratedQuota self-service absent
CostPer rule/request———

GuardDuty — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer volume/ingested log———

CloudTrail — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostFree tier + per-eventSameSame—

AWS Config — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer configuration item———

DevOps toolchain

CloudFormation — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasNot listedNot listedNot listed—
CostFree (resources billed)SameSame—

ECR — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas23 items22, values identical221 fewer item
CostPer storage/transfer———

ECR covers your private images. Pulling public Docker Hub base images on Mainland China runners is a separate failure mode — see Docker Hub in China — mirrors and private registries.

CodePipeline — parity (and its quota plane IS integrated — an earlier draft of our notes said otherwise; corrected against measured data)

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas29 items29 (20 not adjustable)29Identical
CostPer pipeline———

CodeBuild — works; quota plane absent, which makes CI concurrency opaque

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas38 itemsNot integratedNot integratedYou cannot read your build-concurrency ceiling from the API
CostPer build minute———

CodeDeploy — works; fleet sizes halved

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas2,000 instances per deployment; 3,000 concurrent1,000 / 1,000SameLarge fleets deploy in batches
CostPer deploymentFreeFree—

Observability

CloudWatch — core parity; RUM out, query throttles lower

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesMetrics / logs / Synthetics / SLO / RUMFirst four work; RUM not offeredSameFront-end monitoring needs a third party
Quotas69 log items; StartQuery 10 TPS55 items; 6 TPSSameLog-query throttle 10→6
CostPer metric/log volume———

X-Ray — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer trace———

Analytics

Athena — concurrent queries one tenth of global

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
Quotas30 items; active DML queries 2005 items; 202010× fewer concurrent queries — dashboards queue
CostPer data scanned———

Glue — parity

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostPer DPU-hour———

EMR — works; release-catalog comparison deferred

Dimensionus-east-1BeijingNingxiaGap
Versions100 release labels measured100100Count truncation suspected — flagged, not claimed
FeaturesWorksWorksWorksParity
QuotasIntegratedIntegratedIntegratedNo differing values
CostEC2 rates in API; EMR surcharge notSameSamePartially computable

QuickSight — API reachable; fresh account not subscribed

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorksAPI reachable (account-level “not subscribed” response is a fresh-account state, not absence)SameSubscribe and go
QuotasIn catalogCatalog entry (Beijing only)No catalog entryCatalog asymmetry unexplained — flagged
CostPer user/session———

SaaS-critical services

SES — deployed in Ningxia only; a region-selection factor

Dimensionus-east-1BeijingNingxiaGap
Versionsn/an/an/a—
FeaturesWorks (production)No endpointWorks — API live, sandbox 200 emails/day measuredEmail-reliant SaaS has exactly one China region option
QuotasIn catalogNo catalog entryCatalog entryCatalog matches deployment
CostPer email———

DMS — works; narrower engine menu

Dimensionus-east-1BeijingNingxiaGap
Versions5 engine versions22Match your source engine against the short list before planning
FeaturesWorksWorksWorksParity
Quotas21 items1616No material differences
CostPer replication instance———

Not deployed in either China region

These endpoints do not exist in Beijing or Ningxia. The replacement path is the actionable content:

ServiceCategoryReplacement path
App RunnerApp platformECS/Fargate + ALB
Amplify HostingApp platformS3 + CloudFront or third-party
PinpointEngagementDomestic SMS/push vendors
AppFlowData integrationCustom ETL
CodeArtifactDevToolsSelf-hosted Nexus/Artifactory with in-country mirroring
Global AcceleratorNetworkCloudFront / third-party CDN for anycast needs
Cognito User PoolsIdentitySelf-built auth or third-party IdP (identity-pool endpoint exists in Beijing — flagged for review)
CloudWatch EvidentlyObservabilityThird-party A/B tooling (also retiring globally)
Incident ManagerObservabilitySSM + custom incident tooling
Cloud ShellToolingLocal CLI / CI runners

Dimension insights

Versions: sync is the norm, and the exceptions fit on one line

Of the 17 engine lines with a comparable version axis, 12 match the latest version in both China regions — including EKS (1.37, with Auto Mode), MSK (36 Kafka versions), and the entire open-source database line, down to Valkey 9.1, an engine that first shipped in 2024. The “China database versions lag” folklore is not supported by measurement.

The exceptions, in full: Aurora PostgreSQL is two minors behind (18.4 vs 18.6); Amazon MQ is the only major-version gap — both RabbitMQ (4.2 vs 4.3) and ActiveMQ (5.18 vs 5.19); Neptune is ahead in both China regions (1.4.8.1 vs 1.4.8.0), the dataset’s only inversion; ElastiCache memcached drops four historical versions in Ningxia. If your workload is version-sensitive, your entire pre-migration version audit is these four lines.

Features: whole families are missing, not thinned

Feature gaps in the China partition are structural absences, not degradation. EC2 has 432–461 instance types against 1,375 — but within any family that exists, every size matches global exactly (g4dn 7/7, g5 8/8). The same pattern repeats at every layer: OpenSearch exists, OpenSearch Serverless does not; API Gateway REST exists, HTTP API does not; CloudFront distributions exist, the entire edge-compute plane (Functions, KV stores, realtime logs, and Lambda@Edge by dependency) does not.

AWS China GPU instances warrant specific attention in architecture reviews: three accelerator chips (T4, A10G, Inferentia) against thirteen in us-east-1, including the 2026-generation B200/B300, and the 7th-generation CPU families (m7i/c7i/r7i) have not landed. Performance-generation-sensitive workloads face a hard ceiling. On the AI stack, the scope note from the introduction applies: no Bedrock, no managed AI APIs, SageMaker self-build with an A10G ceiling.

Quotas: rate limits, not resource limits

This dimension produced the numbers most likely to change a migration plan. Resource-class quotas (vCPUs, DB instances, tables) are mostly partition-consistent at the default layer — but the rate-class quotas run 10–25× below global: EventBridge invocations at 750 vs 18,750 TPS, SNS topic operations at 120 vs 3,000 TPS, KMS symmetric crypto at 10k vs 100k TPS, Athena concurrent queries at 20 vs 200. These limits bind CI/CD pipelines, event backbones, and BI dashboards on launch day, not at scale.

Default-layer rate limits in AWS China as a multiple of global, log scale. EventBridge and SNS run 25 times tighter, KMS and Athena 10 times, and the dashed line marks global parity.

Three structural findings around them:

  1. The quota management plane is itself gapped. Global exposes 320 services through self-service quotas; China exposes 119/111 — and the missing set includes EKS, API Gateway, CodeBuild, WAF, and EFS. For those, the ceiling is real but unreadable from the API.
  2. “Not adjustable” is the default posture. 94% of Beijing’s EC2 quota items carry no increase path. You design around them; you don’t file tickets.
  3. Intra-China asymmetry is real and material. RDS instances default to 20 in Beijing vs 40 in Ningxia; the S3 object-size quota is reported under two conventions (5 TB vs 48.83 TB — pending verification); SES exists only in Ningxia. “Which China region” is a first-order architecture decision, not a footnote.

Cost: Beijing premium, Ningxia discount

AWS China pricing divides cleanly by region (CNY list prices converted at ≈7.1). Beijing runs +23–49% over us-east-1 on compute (m6i.xlarge +49%, c6i +23%), adds a 153% premium on ALB LCU-hours, and charges +41% for Transit Gateway attachments. Ningxia, meanwhile, prices at or below us-east-1 almost across the board: m5.xlarge −1%, GPU g5.12xlarge −6%, EBS gp3 −6%, ALB −37%, Lambda at exact parity, ElastiCache at parity. Public IPv4 addresses are cheaper in both (−7%).

Unit prices in Beijing and Ningxia as a percent difference from us-east-1, two panels. Beijing premiums are led by ALB LCU-hours at +153%; Ningxia sits mostly at parity or below, down to −35% on t3.medium.

For budget modeling: EC2, EBS, S3, ALB, ElastiCache, Lambda, EKS, TGW, and PrivateLink are all API-priced for China and captured throughout this article. The Cost rows elsewhere in the matrix that carry dashes were simply not captured by this benchmark — not claimed absent. What the China Pricing API genuinely lacks is a short list:

The Pricing API blind-spot list

SurfaceStatus in the China Pricing APIWhere to price it
RDS compute — MySQL, PostgreSQL, MariaDBVerified absent. The China catalog carries storage and license SKUs but no instance billing rowsOfficial AWS China price list
Aurora compute — MySQL, PostgreSQLVerified absent. Same catalog gapOfficial AWS China price list
NAT GatewaySKU not located in the catalog pull — open item, listed in “What we could not pin down”Official AWS China price list
Every other priced line in this articlePresent and automatable. S3, ELB, ElastiCache, the VPC family, EKS, Lambda, and EBS (under the AmazonEC2 service code) were all read via full-catalog pulls with client-side SKU matchingPricing API

Two practical notes for FinOps automation. The catalog pull must be location-based — server-side usagetype filters silently return nothing for China regions — and EBS hides under the AmazonEC2 service code with CNN1-/CNW1-prefixed usagetypes. Once those two conventions are handled, everything in the bottom row scripts cleanly; the database lines do not, as of data version v2026-10.

The compute bill: EC2 and EKS

Every line measured at Linux / shared tenancy / on-demand, shown in USD (China CNY list prices converted at ≈7.1).

Compute line itemus-east-1BeijingNingxia
EC2 t3.medium (burstable)$0.0416$0.037 (−11%)$0.027 (−35%)
EC2 t4g.medium (Graviton burstable)$0.0336$0.029 (−15%)$0.021 (−38%)
EC2 m5.xlarge (general)$0.192$0.285 (+48%)$0.191 (−1%)
EC2 m6i.xlarge (general)$0.192$0.285 (+49%)$0.191 (−1%)
EC2 m7i.xlarge (general)$0.202not offerednot offered
EC2 c6i.xlarge (compute)$0.170$0.208 (+23%)$0.139 (−18%)
EC2 c7i.xlarge (compute)$0.179not offerednot offered
EC2 r6i.xlarge (memory)$0.252$0.343 (+36%)$0.249 (−1%)
EC2 r7i.xlarge (memory)$0.265not offerednot offered
EC2 g4dn.xlarge (1× T4)$0.526$0.736 (+40%)$0.523 (−1%)
EC2 g5.12xlarge (4× A10G)$5.672$7.56 (+33%)$5.32 (−6%)
EKS control plane (per cluster)$0.10/hr$0.097 (≈parity)$0.097 (≈parity)
EKS Fargate (per vCPU-hr)$0.0405$0.051 (+26%)$0.051 (+26%)

Note the inversion: standard families carry a Beijing premium of +23–49%, but burstable t3/t4g run 11–38% below us-east-1 in both China regions, making small dev/test workloads among the few things genuinely cheaper in China. The 7th-generation rows express the generation gap from the features section as money.

The database bill: managed vs self-hosted

Managed database pricing splits into what the API exposes and what it doesn’t. RDS and Aurora compute SKUs are the one blind spot — us-east-1 baselines below are API-measured; China prices for those four rows must come from the official China price pages or the Amazon Pricing Calculator (China region) for interactive estimates. Everything else is measured.

Database line itemus-east-1BeijingNingxia
RDS MySQL db.m5.large (Single-AZ)$0.171/hrofficial price pagesofficial price pages
RDS MySQL db.r5.xlarge (Single-AZ)$0.480/hrofficial price pagesofficial price pages
Aurora MySQL db.r5.xlarge$0.580/hrofficial price pagesofficial price pages
Aurora PostgreSQL db.r5.xlarge$0.754/hrofficial price pagesofficial price pages
ElastiCache Redis cache.r5.large$0.216/hr$0.235 (+9%)$0.213 (−1%)
Self-hosted MySQL/PostgreSQL on r6i.xlarge$0.252/hr$0.343 (+36%)$0.249 (−1%)
Self-hosted Redis/Kafka node on m6i.xlarge$0.192/hr$0.285 (+49%)$0.191 (−1%)

The self-hosted rows quantify the fallback where a managed price is not API-readable: the floor is an EC2 node that prices exactly. In Ningxia that floor sits at −1% vs us-east-1.

The storage bill: S3 classes and EBS volume types

S3 storage classes (GB-month; the 2026 price catalog renames Standard to “General Purpose” and Glacier tiers to “Archive” — old names shown for readability):

S3 storage classus-east-1BeijingNingxia
Standard (General Purpose)$0.023$0.027 (+19%)$0.025 (+8%)
Standard-IA (Infrequent Access)$0.0125$0.019 (+51%)$0.015 (+16%)
Glacier Instant Retrieval$0.0040$0.00465 (+18%)$0.00423 (+6%)
Glacier Flexible Retrieval$0.0036$0.00423 (+18%)$0.0038 (+6%)
Reduced Redundancy$0.024$0.024 (−1%)$0.024 (−1%)

EBS volume types (GB-month, full ladder):

EBS volume typeus-east-1BeijingNingxia
gp3 (general SSD)$0.080$0.084 (+5%)$0.075 (−6%)
gp2 (legacy general SSD)$0.100$0.105 (+5%)$0.094 (−6%)
io1 / io2 (provisioned IOPS)$0.125$0.121 (−3%)$0.108 (−14%)
st1 (throughput HDD)$0.045$0.047 (+5%)$0.042 (−6%)
sc1 (cold HDD)$0.015$0.016 (+5%)$0.014 (−6%)

The storage pattern is consistent: Beijing carries roughly +5% on block storage and +8–51% on object storage by class, while Ningxia lands at or below us-east-1 on every row; provisioned-IOPS SSD is 14% cheaper there. Archive classes carry a +6–18% premium over us-east-1 — modest, but log-retention and backup-heavy SaaS should still re-run the math before porting a retention policy.

One claim withdrawn during review: an earlier draft flagged the Beijing g5 ladder as inverted because g5.12xlarge (4× A10G, $7.55) lists above g5.16xlarge (1× A10G, $5.46). Sorting by GPU count shows this mirrors the global structure exactly — us-east-1 prices the same pair at $5.67 vs $4.10. The ladder is GPU-count-driven; there is no anomaly.

Four conclusions from 66 services

Read across the four dimensions and four structural conclusions emerge:

1. The catalog was never the problem. Of the 76 service surfaces a complex SaaS would touch, 66 are deployed, healthy, and measured below, and 10 have no endpoint in either region; on the version axis, 12 of 17 engine lines match global’s latest in both regions. A team that stops at “is it listed?” predicts failure where measurement shows parity, and misses it where it actually occurs.

2. The gaps are in rate and management planes. The gaps that bite are EventBridge at 4% of global TPS, KMS at 10%, Athena at 10%, quota APIs that cannot answer for EKS and CodeBuild, and an edge-compute plane that doesn’t exist. None of these appear in a catalog; they appear in production.

3. Beijing and Ningxia are different products. The intra-China asymmetries (RDS 20 vs 40, SES Ningxia-only, S3 replication destinations 2 vs 28, ALB LCU +153% vs −0%, EBS +5% vs −6%) mean region selection is an architecture decision with quota, cost, and feature consequences — made once, at the start. For cost-sensitive, latency-tolerant workloads the data says Ningxia; for anything needing SES or lower latency to North China, Beijing — with its premium — is the only option.

4. A fresh account is a known quantity. Because this baseline was measured on day-zero accounts, a new product team can read every default above as their own starting position — and the two-layer quota method in our field note tells them how to keep reading it correctly as they grow.

Decision rules, condensed:

Your workloadThe measured answer
Data-layer lift-and-shiftMigrate — 12/17 engine lines at the latest in both regions; check Aurora PG (−2 minors) and MQ (−1 major)
Event-driven / high-fanoutRe-rate the backbone: EventBridge and SNS at 1/25th global TPS
CI/CD at fleet scaleCodeBuild quotas unreadable; CodeDeploy 1,000 instances per deployment
BI-heavyAthena 20 concurrent queries; budget for queues or caching
Crypto-heavyKMS 10k TPS symmetric; cache envelope keys
Edge-dependentNo Functions/Lambda@Edge; Ningxia CloudFront retires 2027-05-31
Email-sending SaaSNingxia or nothing
Cost-optimizedNingxia, by the numbers

What we could not pin down

Published, not hidden. Status vocabulary: verified (double-checked), needs-console-review (API evidence is strong but the console must confirm), inconclusive (this method cannot resolve it).

#ItemStatus
1SES Ningxia “deployed” verdict (API 200 + sandbox quota; contradicts common belief)needs-console-review
2Cognito identity-pool asymmetry (Beijing yes, Ningxia no endpoint)needs-console-review
3S3 object-size quota convention: 5 TB (Beijing) vs 48.83 TB (Ningxia, us-east-1) under one quota nameneeds-console-review
4RDS default instances Beijing 20 vs Ningxia 40needs-console-review
5Lambda runtimes (no standalone read-only API)inconclusive
6EMR release-catalog counts (possible pagination truncation)inconclusive
7NAT Gateway price SKU not located in the VPC cataloginconclusive
8ElastiCache cache.m5.large absent from China catalog (m6g likely substitute)needs-console-review
9PrivateLink / TGW quota codes not separately listedneeds-console-review

Appendix: reproduce every table

The claim “all commands re-runnable” should be checkable, not trusted. Every command below is read-only (describe / list / get), costs nothing, and produces rows in this article. China-region calls need a China-partition profile; us-east-1 calls need a global one.

Command list (click to expand)
# Quota catalog coverage — the 320 vs 119/111 table
aws service-quotas list-services --region cn-north-1

# Default-layer quotas per service — the layer every quota number here uses
aws service-quotas list-aws-default-service-quotas --service-code ec2 --region cn-north-1

# Two-layer quota read — the field-note method (applied vs default)
aws service-quotas get-service-quota --service-code lambda --quota-code L-B99A9384 --region cn-north-1
aws service-quotas get-aws-default-service-quota --service-code lambda --quota-code L-B99A9384 --region cn-north-1

# Engine versions — the version tables
aws eks describe-cluster-versions --region cn-north-1
aws rds describe-db-engine-versions --engine aurora-postgresql --region cn-north-1
aws elasticache describe-cache-engine-versions --region cn-northwest-1
aws kafka list-kafka-versions --region cn-north-1
aws mq describe-broker-engine-types --region cn-north-1
aws opensearch list-versions --region cn-north-1

# Compute breadth and GPU inventory
aws ec2 describe-instance-type-offerings --region cn-north-1
aws ec2 describe-instance-types --instance-types g5.12xlarge --region cn-north-1   # GpuInfo

# Feature presence — one minimal read probe per service
aws sqs list-queues --region cn-north-1
aws apigatewayv2 get-apis --region cn-north-1          # HTTP API: fails in China regions, by design of the probe
aws sesv2 get-account --region cn-northwest-1          # SES: Ningxia responds, Beijing has no endpoint

# Prices — Price List API, always queried via the us-east-1 pricing endpoint
aws pricing get-products --service-code AmazonEC2 \
  --filters 'Type=TERM_MATCH,Field=regionCode,Value=cn-north-1' 'Type=TERM_MATCH,Field=instanceType,Value=m5.xlarge'
aws pricing get-products --service-code awselb \
  --filters 'Type=TERM_MATCH,Field=location,Value=China (Beijing)'   # full catalog pull; match SKUs client-side

Notes: quota names and engine version lists are joined across the three regions on identical keys, and the dataset’s source_api column names the exact call behind every row. RDS and Aurora China prices cannot be produced this way — that is the documented API blind spot.

Data version, citation, and requesting a test

Data version: v2026-10 (measured 2026-10-11/12). No fixed re-test cadence is promised — most of these measurements are one-shot by design. Re-measurements and extensions increment the version and land in the changelog.

Download the dataset. Every quantitative row in this article — quota values, engine versions, probe statuses, and unit prices, 224 rows with the source API named on each — is exportable as machine-readable data:

  • JSON (aws-china-partition-benchmark-v2026-10.json, ~60 KB)
  • CSV (long format: service, dimension, metric, region columns, source API)

Both carry the same metadata block as this article (method, dates, conventions, withdrawn claims) and are served under CC BY 4.0 with attribution to Chinaready.

Changelog

  • v2026-10 — 2026-10-12 — Initial release: 66 services × 4 dimensions × 3 regions; quota catalog diff (320/119/111); per-service default-layer quota alignment; unit prices for 10+ cost lines; RDS/Aurora identified as the Pricing API blind spot for China.

Cite this dataset

Chinaready. "AWS China vs Global: Partition Differences Measured Across 66 Services."
Chinaready Insights, four-dimension partition benchmark, data version v2026-10, 2026-10-12.
https://chinaready.co/insights/aws-china-partition-benchmark/
<cite>Chinaready. <a href="https://chinaready.co/insights/aws-china-partition-benchmark/">AWS China vs Global: Partition Differences Measured Across 66 Services</a>. Chinaready Insights, four-dimension partition benchmark, data version v2026-10, 2026-10-12.</cite>

Attribution, trademarks, and disclaimer

Disclaimer. The measurements and conclusions in this article are an independent, user-side evaluation conducted by Chinaready. They do not represent the official position of Amazon Web Services, Inc. or the operators of the AWS China regions — this is a customer’s benchmark, not an AWS test. All AWS product names and trademarks are the property of AWS. For authoritative answers on AWS products, pricing, or quotas, use the official sources listed at the top of this article or contact AWS China through its official channels.

Attribution and copyright. The dataset downloads (JSON and CSV) are licensed under CC BY 4.0 with attribution to Chinaready. The article text is © 2026 Chinaready; quote or excerpt it with a link to this page.

Request a test

This dataset covers what a migration review needs, not everything. If a service, region pair, or dimension you care about is missing or stale, request a test. Two ways in:

  1. Use the contact form — tell us what you need measured and why it matters to your architecture.
  2. Email [email protected] with your test request.

Selected requests get measured and folded into the next data version with a changelog entry. Requests that map to a real architecture question tend to get picked first.

Where to go next

Frequently asked questions

Can my team reproduce these measurements?
Yes. Every table in this article names the API behind it, and all of them are read-only (describe, list, get) against your own account pair — a China-partition account and a global account. Expect zero resource creation and zero spend. Re-run the same commands and compare against this data version; drift over time is itself useful signal.
Does not being integrated with Service Quotas mean EKS is unavailable in China?
No. EKS is deployed and current in both Beijing and Ningxia — cluster versions 1.32–1.37 match us-east-1, and Auto Mode is available. Not integrated refers only to the quota self-service API. EKS limits cannot be queried or raised through Service Quotas in the China partition, so capacity planning relies on AWS documentation and support cases.
Are these quota values what my account will actually receive?
They are default-layer values, which is exactly what a brand-new account receives. Accounts with usage history can hold different applied values, and a quota that was never initialized is invisible to the applied-value API even when it exists. That is why we read both API layers for every service before drawing a conclusion.
How current is this data, and will it be refreshed?
All measurements are dated 2026-10-11/12 and versioned as v2026-10. No fixed re-test cadence is promised — most of these checks are one-shot by design. Revisions and extensions increment the data version and are recorded in the changelog; readers can request specific re-tests through the contact channels at the end of this article.
Why do Beijing and Ningxia differ from each other?
The two China regions are operated by different local partners and priced independently, and several defaults diverge. RDS instances default to 20 in Beijing versus 40 in Ningxia, the S3 object-size quota is reported inconsistently across regions (5 TB versus 48.83 TB — an API-convention discrepancy pending verification, since the documented single-object limit is 5 TB in every region), and SES is deployed only in Ningxia. Region choice is therefore an architecture decision with quota, cost, and feature consequences — not a formality.

Tell us where you are stuck in China.

Share your product, stack, and timeline — we will point you to the next concrete step.