Log in Start Free
Start Free

QDRANT CLOUD

Production-Ready Resilience, Managed on Qdrant Cloud

Go from single-node to fully replicated high availability in the Console. Add Multi-AZ on the Premium tier for zone-level resilience and an uptime SLA of up to 99.95%, and the cluster stays available through a full availability-zone outage.

Start Free Read the Docs

Need four or five nines with contractual guarantees? Talk to our team about Enterprise and Private Cloud.

Create a Cluster Create a Cluster

How Resilience Works
on Qdrant Cloud

Run at least 3 nodes with replication factor 2 or higher and the capabilities below apply to your cluster.

Configure a Replicated Cluster (Docs)
Copy
Replication

Equal copies of every shard, kept across your nodes.

File
Failover

Searches and writes continue when a node goes down.

Refresh
Rolling Upgrades

Updates move through one node at a time while the rest serve traffic.

You set it, Qdrant runs it. You choose the replication factor, the platform maintains the replicas.

CLUSTER CONFIGURATION

Configure the Right Resilience for Your Workload

Replication and node count give you high availability. Multi-AZ is the zone-resilience upgrade on top, for teams that need to survive a full availability-zone outage. These three properties define your cluster's resilience on Qdrant Cloud:

Replication Factor
Replication factor

What Qdrant Cloud Does Keeps equal copies of every shard across your nodes

What it Means Searches and writes continue when a node goes down.

Node Count
Node Count

What Qdrant Cloud Does Distributes shard replicas across more nodes.

What it Means More headroom for your cluster to stay healthy.

Multi-AZ
Multi-AZ

What Qdrant Cloud Does Spreads those nodes across three availability zones.

What it Means Your cluster stays up across a full zone outage.

Boxes
Multi-AZ
  • Enabled at cluster creation
  • Available on the Premium tier
  • Changes where your replicas are placed, not how many
  • Without it, a multi-node cluster keeps all replicas in one zone
Read Documentation

CAPABILITIES

These capabilities apply to replicated collections.
See cluster configuration above for the node count and replication factor requirements.

Configure a Cluster (Docs)

What Qdrant Cloud Manages, By Tier

Start Free Talk to Engineering
Copy ALL REPLICATED CLUSTERS

Automatic Failover

Mechanics Qdrant Cloud runs health checks across nodes and routes around unhealthy nodes, so queries continue from healthy nodes. Every replica is equal, so there is no primary to promote.

Guarantees Qdrant Cloud detects failures and keeps serving traffic from healthy nodes. No client-side changes required.

Set Up a Replicated Cluster (Docs)
Cloud cluster UI Cloud cluster UI
Copy ALL REPLICATED CLUSTERS

Zero-Downtime Upgrades

Mechanics When a new version is available, you choose when to upgrade. Qdrant Cloud upgrades in a rolling fashion. If you are several versions behind, the required intermediate updates run automatically.

Guarantees Multi-node clusters where all collections have replication factor 2 or higher stay fully available throughout the upgrade.

Update Your Cluster (Docs)
Cluster overview UI Cluster overview UI
Dollar ALL PAID CLUSTERS

Backups and Disaster Recovery

Mechanics Schedule backups, set days of retention, or take an on-demand backup any time. Backups restore into the same cluster or a new one.

Guarantees Restore your cluster to the exact state of any backup within your retention window, including its configuration. No recovery scripts needed, so your team can focus on getting back online.

Back Up Your Cluster (Docs)
Backups UI Backups UI

Honest Note Backups protect against data loss; they are not a fast-failover substitute. For low recovery time, run replicated and Multi-AZ. For planned data-center switchovers or full region loss, talk to us about an active-active pattern.

Phone TIER-DEPENDENT

Engineering Support

Mechanics Premium support covers production incidents 24/7, with severity-based response-time SLAs. Standard covers business hours.

Guarantees Faster response times and around-the-clock coverage on Premium, with SLAs defined per severity level in the Qdrant Cloud SLA.

See Support SLAs
Trophy PREMIUM-TIER

Multi-AZ Placement

Mechanics Qdrant Cloud places your shard replicas across three availability zones, so each shard has a replica in a different zone. Traffic is routed between zones automatically, so the cluster stays available if one zone goes down. Enable Multi-AZ Deployment when you create the cluster.

Guarantees An uptime SLA of up to 99.95% on the Premium tier with Multi-AZ enabled. Traffic reroutes across zones automatically, with no client-side changes.

Start Free

DEPLOYMENT MODELS

What Resilience You Get on Each Deployment Model

Managed Cloud
Hybrid Private
Self-Hosted OSS
Replication Factor Set in Console
RAutomatic Failover with RF 2+
Multi-A-Z
Premium, Enabled at creation
Your Kubernetes placement
Customer operated, no topology aware shard distribution
Zero-downtime Upgrades
Replicated
Replicated
Manual
Auto Rebalance
Backups
Scheduled
+ on-demand (all paid)
Scheduled
+ on-demand (all paid)
Snapshot API
Uptime SLA
99.5% / 99.9%
Premium / 99.95%
Premium + Multi-AZ
Your Infrastructure

The capabilities above apply to Qdrant Cloud.
The same resilience primitives run across every deployment model.

Compare Deployment Models (Docs)

RESILIENCE IN PRODUCTION

How Teams Maintain Uptime at Scale

Sapu migrated from self-hosted to Qdrant Cloud Premium, indexing 28 million PubMed abstracts in a single collection.

Read the Case Study
SAPU logo

FAQs

What's the uptime SLA?
Up to 99.95% uptime on the Premium tier with Multi-AZ enabled. Full SLA terms are in the Qdrant Cloud SLA.
What's the difference between Standard, Premium, and Premium Multi-AZ SLAs?
Standard offers a 99.5% uptime SLA. Premium raises it to 99.9%. Premium with Multi-AZ enabled reaches 99.95%. Each tier also differs on support coverage and response times.
How is uptime measured?
Please refer to our Qdrant Cloud SLA.
What if I need a higher SLA than 99.95%?
Talk to us. We arrange bespoke SLAs for specific deployments.
What happens during a zone failure?
On Multi-AZ clusters, reads and writes continue from the surviving zones automatically, and you do not need to take any action.
Do I need to change my client code for failover?
Failover happens server-side, so all clients behave the same way. For transient errors during a failover, retry with backoff, the standard production pattern.
Is Multi-AZ the same as failover?
No. Multi-AZ is continuous cross-zone replication that keeps the cluster available across availability zones. Node-level failover, where unhealthy nodes drop from rotation, is a separate mechanism.
Does a multi-node cluster spread across availability zones automatically?
No. A multi-node cluster gives you replication across nodes, but those nodes can sit in the same availability zone. Zone distribution only happens when you enable Multi-AZ at cluster creation. Replication factor and zone placement are two independent properties.
What do I need to get these resilience capabilities?
Run replicated collections. Your cluster should have at least 3 nodes, and each collection should have a replication factor of at least 2 (3 recommended for Multi-AZ). On Qdrant Cloud, Qdrant adds or drops shard replicas automatically to match the replication factor you set.
How do I enable Multi-AZ?
Check the Multi-AZ Deployment checkbox when you create the cluster. Multi-AZ clusters need a minimum of 3 nodes and scale in multiples of 3. Multi-AZ is available on the Premium tier.
Can I add Multi-AZ to an existing cluster?
No. Multi-AZ can't be added to an existing cluster. To move an existing workload onto Multi-AZ, create a new Multi-AZ cluster and migrate your data. Talk to engineering if you need help.
Is Qdrant replication primary/secondary?
There is no primary, no leader, and no write hot spot. Every replica is equal and any node accepts reads and writes. You set a replication factor and Qdrant keeps that many copies of each shard across your nodes. Write consistency is governed by the consistency factor you configure.
Are backups automatic?
Backups are under your control. Choose a schedule in the Console Backups tab and set how long to keep each one with days of retention, or take an on-demand backup any time. You only pay for the backups you configure.
Can I restore a backup to a different cluster?
Yes. Restore a backup into the same cluster to revert it, or restore into a new cluster.
What does a restore actually recover to?
A restore returns your cluster to the exact state captured in the backup, including its CPU, memory, node count, and Qdrant version at that time. Any changes made after the backup date are lost. The cluster is unavailable while the restore is in progress, and restore time depends on the size of your data. For recovery-time guidance on your workload, talk to engineering.
Are upgrades really zero downtime?
For collections with a replication factor of at least 2, yes. Qdrant Cloud uses a rolling restart, updating nodes one at a time while peers serve traffic. If all collections have replication factor of 1, it uses a parallel restart, which causes a short downtime.
Do I control when upgrades happen?
Yes. When a new version is available, Qdrant Cloud shows an update notification on the Cluster Details page. Choose the version and click Update. You can update at any time, and if you are several versions behind, Qdrant Cloud performs the required intermediate updates for you.
Does my cluster get all of these capabilities, or just some?
Run replicated collections (3 or more nodes, replication factor 2 or more) to get automatic failover, scheduled backups, and zero-downtime upgrades. Multi-AZ replication across three availability zones, and the 99.95% uptime SLA, are on the Premium tier.
How do you handle data residency?
Choose your data-center region when you create the cluster, across AWS, Azure, and Google Cloud. For physical isolation, see Private and Hybrid Cloud.

Build on Qdrant Cloud

Replication, failover, and scheduled backups on every replicated cluster.

Start Free Talk to Engineering
Rocket flying over globe illustration