Can an app profile use both multi-cluster and single-cluster routing?

No, they're mutually exclusive on the same app profile; you set exactly one. If you need both automatic failover for most traffic and single-cluster guarantees for a subset, create two app profiles and point each workload at the one matching its consistency needs.

What does Data Boost isolation actually change compared to standard isolation?

Standard isolation shares the cluster's own provisioned nodes, so a LOW-priority read can still queue behind HIGH-priority traffic under load. Data Boost runs reads on separate serverless compute that Google manages outside your provisioned node count entirely, so a heavy analytical scan literally cannot slow down your serving path, at the cost of it being read-only and billed separately.

Why did enabling allowTransactionalWrites not help my read-modify-write consistency?

It only has an effect combined with singleClusterRouting. Setting allowTransactionalWrites: true under a multiClusterRoutingUseAny profile does nothing, because the guarantee Bigtable provides for check-and-mutate/read-modify-write atomicity depends on all requests landing on the same cluster, which multi-cluster routing by design doesn't promise.

GCP Bigtable App Profile Builder

StorageGCP

Build Bigtable app profile configurations with multi-cluster routing, isolation priorities, and Data Boost.

Bigtable App Profiles Configuration

Build Bigtable app profile configurations with multi-cluster routing, single-cluster routing, isolation priorities, and Data Boost.

Required Fields

namedescription

{
  "name": "projects/my-project/instances/prod-bigtable/appProfiles/analytics-reader",
  "description": "App profile for analytics read workloads with multi-cluster routing",
  "multiClusterRoutingUseAny": {
    "clusterIds": [
      "prod-cluster-us-central1",
      "prod-cluster-us-east1"
    ]
  },
  "singleClusterRouting": null,
  "standardIsolation": {
    "priority": "PRIORITY_LOW"
  },
  "dataBoostIsolationReadOnly": {
    "computeBillingOwner": "HOST_PAYS"
  },
  "etag": "",
  "alternativeProfiles": [
    {
      "name": "projects/my-project/instances/prod-bigtable/appProfiles/realtime-writer",
      "description": "Single-cluster routing for low-latency writes",
      "singleClusterRouting": {
        "clusterId": "prod-cluster-us-central1",
        "allowTransactionalWrites": true
      },
      "standardIsolation": {
        "priority": "PRIORITY_HIGH"
      }
    },
    {
      "name": "projects/my-project/instances/prod-bigtable/appProfiles/batch-export",
      "description": "Data Boost profile for batch export jobs",
      "multiClusterRoutingUseAny": {
        "clusterIds": []
      },
      "dataBoostIsolationReadOnly": {
        "computeBillingOwner": "HOST_PAYS"
      }
    }
  ]
}

Generated Output

Output will appear here...

See It in Action

A payments team notices occasional double-charges under load and traces it to a Bigtable read-modify-write balance update running through the default app profile, which was configured with multiClusterRoutingUseAny — meaning consecutive requests for the same row could land on different clusters and race each other. They use the builder to add a dedicated singleClusterRouting profile with allowTransactionalWrites: true pinned to the primary cluster, repoint the balance-update path at it, and leave the original multi-cluster profile in place for the read-heavy dashboard queries that don't need same-cluster guarantees.

What This Tool Does

Build Bigtable app profile configurations that control how different applications route requests to a multi-cluster instance. multiClusterRoutingUseAny and singleClusterRouting are mutually exclusive: pick multi-cluster for automatic failover across regions, or single-cluster when you need allowTransactionalWrites for read-modify-write operations, which Bigtable only guarantees are atomic within one cluster. standardIsolation.priority (LOW/MEDIUM/HIGH) lets you keep a noisy batch-read app profile from starving a latency-sensitive serving app profile that shares the same cluster's fixed node capacity.

Technical Details

The builder only requires name and description to resolve, then leaves routing mode, isolation, and Data Boost settings as free-form JSON matching Bigtable's AppProfile resource, since which routing block you include (multiClusterRoutingUseAny vs singleClusterRouting) is a semantic choice the tool can't validate — it can confirm the JSON is well-formed but not that you picked the routing mode your workload's consistency requirements actually need.

Common Use Cases

1Splitting a shared Bigtable cluster into a HIGH-priority app profile for the user-facing serving path and a LOW-priority app profile for a nightly batch export, so the batch job can't burn through cluster capacity during peak traffic.
2Adding singleClusterRouting with allowTransactionalWrites for a service doing check-and-mutate operations, after discovering multi-cluster routing doesn't provide the same-cluster guarantee those operations depend on.
3Creating a dataBoostIsolationReadOnly profile with computeBillingOwner HOST_PAYS for an analytics team's ad-hoc queries, so their serverless read compute never contends with the provisioned cluster nodes at all.
4Auditing alternativeProfiles before a cluster resize, since a profile pinned to a specific clusterId that gets removed during a topology change will start failing requests instead of failing over.

Common Questions

Can an app profile use both multi-cluster and single-cluster routing?: No, they're mutually exclusive on the same app profile; you set exactly one. If you need both automatic failover for most traffic and single-cluster guarantees for a subset, create two app profiles and point each workload at the one matching its consistency needs.
What does Data Boost isolation actually change compared to standard isolation?: Standard isolation shares the cluster's own provisioned nodes, so a LOW-priority read can still queue behind HIGH-priority traffic under load. Data Boost runs reads on separate serverless compute that Google manages outside your provisioned node count entirely, so a heavy analytical scan literally cannot slow down your serving path, at the cost of it being read-only and billed separately.
Why did enabling allowTransactionalWrites not help my read-modify-write consistency?: It only has an effect combined with singleClusterRouting. Setting allowTransactionalWrites: true under a multiClusterRoutingUseAny profile does nothing, because the guarantee Bigtable provides for check-and-mutate/read-modify-write atomicity depends on all requests landing on the same cluster, which multi-cluster routing by design doesn't promise.

Expert Tips

TIP

A cluster ID referenced by a single-cluster-routing app profile that gets deleted (e.g., during a region migration) breaks that profile's requests outright — Bigtable doesn't silently reroute it, unlike a multi-cluster profile which just drops the dead cluster from its candidate set.

TIP

Data Boost is read-only; a write sent through a dataBoostIsolationReadOnly profile is rejected, so it can't be a drop-in replacement for a standard app profile used by any workload that also writes.

TIP

priority is evaluated only under contention — on an underutilized cluster, LOW and HIGH priority profiles perform identically, so priority misconfigurations often go unnoticed until the first real traffic spike.

Related Learning Guides

GCS Storage Classes & Lifecycle22 min read
Memorystore for Redis Guide24 min read

Was this tool helpful?

Disclaimer: This tool runs entirely in your browser. No data is sent to our servers. Always verify outputs before using them in production. AWS, Azure, and GCP are trademarks of their respective owners.