path-templates · Educational

Cloud egress: Cross-AZ vs Cross-Region Data Transfer

Cross-AZ (cross-zone) chatter and cross-region / inter-region transfer are different meters from Internet DT-out. Batch-21 thickens the worked stack with a numbered InterZone / ELB / DR / CRR table and FAQ that routes LB meters to the ELB spoke. HA often burns more on cross-AZ than on customer downloads — separate the lines; when bytes are S3 CRR, use the S3 CRR egress spoke. Decode InterZone / Inter-AZ labels on the Cost Explorer DT decoder, and keep ALB/NLB cross-zone load balancing on the ELB cross-zone spoke when the LB product meters separately from instance-to-instance cross-AZ.

Updated 2026-09-10

High availability often burns more on cross-AZ chatter than on customer downloads — thicken InterZone, ELB, DR, and CRR as separate Unknown cells.

Path matrix: cross-AZ vs cross-region data-transfer meters (educational)
Meter	Cross-AZ	Cross-region	Watch
Typical trigger	LB→instance, DB replica, chatty microservices	DR replication, multi-region reads	Architecture diagrams
Relative $/GB (educational)	Lower than inter-region (verify)	Higher tiers common	Provider + product
Same-AZ baseline	Often $0 between resources	N/A	Pin locality
User-facing?	Usually internal	Sometimes user geo routing	Don't blend with Internet egress
Control lever	Fewer cross-AZ hops; caching	Async replicate; regionalize	Product constraints

Provider source footnotes

AWS EC2 on-demand data transfer — as-of 2026-09-09 Unknown until verified
GCP network pricing — as-of 2026-09-09 Unknown until verified
Azure bandwidth pricing — as-of 2026-09-09 Unknown until verified

Dated starting points only — Unknown until verified means no invented $/GB. PipeToll does not scrape private consoles or guarantee figures.

Path-class vocabulary: path template, cross-AZ (cross-zone) transfer, cross-region transfer, egress (DT-out), ELB cross-zone, S3 CRR, NAT Gateway processing, free-tier cliff, FinOps-lite, and Unknown until verified. Treat cross-AZ and cross-region as internal path templates distinct from Internet egress. PipeToll’s FinOps-lite habit is to name the path before comparing any $/GB cell.

Path templates to model

Load balancer → instances across AZs — request/response bytes may bill as cross-AZ or as ELB cross-zone depending on product.
Primary DB → standby in another AZ — replication chatter is often invisible in “user egress” dashboards.
Mesh sidecars east-west — chatty RPCs multiply GB even when payloads look small.
Async cross-region replica — DR/backup path (cross-region); when bytes are S3 bucket→bucket use S3 CRR.
Same-AZ pinned pairs — baseline that is often $0 between resources; use as control. NAT hairpins across AZs still stack NAT processing separately.

Common traps

Cross-AZ blended with Internet — One spreadsheet line for “data transfer” hides the HA tax. Keep cross-AZ / InterZone off Internet DT-out; decode labels with the CE decoder.
“We only pay for downloads” — Replica streams, mesh chatter, and sync traffic can exceed user-facing egress. Model east-west Mode A separately from client Internet.
Internet $/GB as the only multi-cloud cell — Comparisons that quote only Internet miss inter-zone SKUs. Sibling depth: AWS vs GCP egress.
Collapse-to-one-AZ as a savings crown — Single-AZ is an availability tradeoff, not a guaranteed savings path. Document RPO/RTO acceptance before cutting HA bytes.
ELB cross-zone pasted into instance InterZone — ALB/NLB cross-zone GB is a different product meter — see ELB cross-zone and ?path=elb-xz.
Every Inter-Region CE label = S3 CRR — Open S3 CRR only when bucket→bucket replication is the driver; otherwise keep a generic cross-region Mode A line.
Scorecrowns from Incomplete — Mark unverified rates Unknown until verified. No guaranteed savings from collapsing AZs.

FinOps checklist

Inventory top cross-AZ talkers (LB, DB, mesh) by bytes — not only by request count; split ELB cross-zone from instance cross-AZ when both appear.
Separate Internet, cross-AZ, and cross-region lines in the cost model; map CE labels via the Cost Explorer decoder.
Run sensitivity with calculator Mode A (?mode=gb&path=cross-az) and, for DR bytes, ?path=cross-region using verified regional rates only.
For HA changes, pair cost deltas with explicit RPO/RTO acceptance — no savings guarantees.
Link findings to glossary: Cross-AZ, Cross-region, and methodology.

FinOps-lite Unknown until verified

Worked example (educational): 8 TB cross-AZ chatter / month

Educational shape: a multi-AZ service ships ~8,000 GB of replica and mesh traffic inside a region (not Internet DT-out).
Price that path alone in Mode A with ?path=cross-az&gb=8000; list $/GB stays Unknown until a dated cite.
Do not blend with Internet egress or with S3 CRR. No savings guarantee if you collapse AZs — availability constraints dominate.

Numbered cross-AZ vs cross-region stack — Unknown rate cells
#	Meter	Volume	Unit rate
1	Instance / mesh cross-AZ chatter	8,000 GB	Unknown InterZone $/GB — Mode A ?path=cross-az
2	ALB/NLB cross-zone (if LB product meters)	as measured	Unknown ELB $/GB — keep off #1
3	Async cross-region DR	1,500 GB	Unknown inter-region $/GB — Mode A ?path=cross-region
4	S3 CRR (only if bucket→bucket)	as measured	Switch to S3 CRR / ?path=s3-crr
5	Internet DT-out (clients)	separate line	Do not blend with #1–#4

Second educational stack (cross-region DR twin): the same month ships 1,500 GB async DR replication to a second region. Keep a separate Mode A line with ?path=cross-region&gb=1500 — row #3 in the table above. If those bytes are S3 bucket→bucket CRR, switch to the S3 CRR egress spoke and ?path=s3-crr instead of a generic inter-region cell. HA sensitivity twin: cut mesh chatter from 8 TB to 4 TB with fewer east-west hops — re-run ?path=cross-az&gb=4000 and document RPO/RTO acceptance; never treat the delta as guaranteed savings. When ALB/NLB cross-zone appears as its own CE line, open ELB cross-zone rather than folding into row #1. Synthetic only — rates Unknown.

Worked-estimate sensitivity (educational)

Hold the 8 TB cross-AZ chatter + 1.5 TB cross-region DR baseline and walk three path-class runs. InterZone $/GB, inter-region $/GB, ELB cross-zone, and Internet DT-out stay Unknown until dated cites — literacy only, not an HA ROI quote.

Run A — Fixed-shape split: 8,000 GB cross-AZ on Mode A (?path=cross-az) plus 1,500 GB async DR on ?path=cross-region. Never blend into one cell; if DR is S3 bucket→bucket, switch to ?path=s3-crr / S3 CRR.
Run B — HA-chatter twin: cut mesh from 8 TB to 4 TB with fewer east-west hops — re-run ?path=cross-az&gb=4000 and document RPO/RTO acceptance. Do not treat the delta as guaranteed savings.
Run C — ELB cross-zone sibling: same month with ALB/NLB cross-zone GB as its own CE line. Keep it off instance InterZone; open ELB cross-zone and ?path=elb-xz. Internet client DT-out stays a separate Mode A line.

Sensitivity runs — Unknown rates (educational)
Run	What changes	Meters to keep visible	Unit rates
A	8 TB cross-AZ + 1.5 TB cross-region split	InterZone · inter-region / S3 CRR (separate Mode A)	Unknown
B	HA-chatter twin (4 TB mesh)	Cross-AZ only · RPO/RTO note	Unknown
C	ELB cross-zone sibling	ELB-XZ off InterZone · Internet DT-out aside	Unknown

Cross-read AWS vs GCP egress, CE decoder, DX / Interconnect, and methodology. Badge every unverified rate Unknown until verified. No savings guarantee from collapsing AZs.

Open calculator · cross-AZ 8 TB Mode A · cross-region 1.5 TB Mode A · HA twin 4 TB Mode A · ELB cross-zone ELB cross-zone CE DT decoder

FAQ on this path

Is cross-AZ the same as Internet egress?

No. Cross-AZ is usually internal regional transfer between zones. Internet egress leaves the provider toward public clients. Keep meters separate in FinOps reviews.

Why can HA cost more than user downloads?

Chatty replicas, synchronous commits, and east-west microservice calls can move more GB than client responses. Measure both.

Does cross-region always cost more per GB?

Often inter-region tiers are higher, but verify per provider and product. Replication schedules and compression change totals.

When is cross-region the same as S3 CRR?

Only when the bytes are S3 bucket→bucket Cross-Region Replication. Generic inter-region / “Regional Data Transfer” talkers may be compute or other products — open the S3 CRR egress spoke when replication rules are the driver.

Should I model ELB cross-zone on this page or the ELB spoke?

Use this page for instance/mesh InterZone vs inter-region literacy. When the load-balancer product meters cross-zone separately, keep that GB on the ELB cross-zone spoke and map CE labels on the CE decoder — never paste ELB GB into the instance cross-AZ cell.

How do I sensitivity-test HA chatter?

Hold client Internet DT-out fixed and vary cross-AZ GB (e.g. 8 TB → 4 TB with fewer east-west hops) in Mode A (?path=cross-az). Pair every cost delta with explicit RPO/RTO acceptance — collapsing AZs is an availability tradeoff, not a savings guarantee.

Is ALB/NLB cross-zone the same as instance cross-AZ?

Not always. Load-balancer cross-zone load balancing can meter on the ELB product separately from instance-to-instance cross-AZ chatter. Keep both lines — see ELB cross-zone and map CE labels on the Cost Explorer decoder.

Trademarks: Amazon Web Services, AWS, Microsoft Azure, Google Cloud Platform, GCP, and Cloudflare are trademarks of their respective owners. PipeToll (Ameti One LLC) is not affiliated with, endorsed by, or sponsored by those companies. Educational comparisons may be outdated — verify current rates with each provider.