This page answers common questions about Lindorm instance operations—including scaling, configuration changes, version upgrades, and restarts—covering how long each operation takes, how it affects your running workloads, and what to check before you begin.
At a glance
The table below summarizes timing and business impact for all major O&M operations. Use it to plan maintenance windows.
| Operation | Duration | Read/write interrupted? | Latency jitter? | Connection may drop? |
|---|---|---|---|---|
| Node scale-out | 10–15 min per node (parallel) | No | Yes | — |
| Storage scale-out | Minutes | No | — | — |
| Enable cold storage | 10–15 min per node (sequential) | No | Yes | Yes |
| Cold storage scale-out | Minutes | No | — | — |
| Configuration upgrade | 10–15 min per node (sequential) | No | Yes | Yes |
| Node scale-in | See formula | No | — | — |
| Configuration downgrade | 10–15 min per node (sequential) | No | Yes | Yes |
| Minor version upgrade — LindormDFS | 10–15 min per node (sequential) | No | — | — |
| Minor version upgrade — LindormTable | 10–15 min per node (sequential) | No | Yes | Yes |
| Instance restart | 10–15 min per node (sequential) | No | Yes | Yes |
For operations where latency jitter or connection drops are possible, configure your client to reconnect automatically, and schedule the operation during off-peak hours if your workload is latency-sensitive.
Scaling out, configuration upgrade, and service activation
How long does it take to scale out the nodes of an instance?
Node scale-out runs in parallel across all nodes being added. Each node takes about 10 to 15 minutes.
What is the impact on my business during the node scale-out?
The instance continues handling reads and writes throughout the scale-out. Some requests may experience temporary latency spikes. Make sure the client is configured to reconnect automatically.
How long does it take to scale out the storage capacity of an instance? What is the impact on my business during the scale-out?
Storage scale-out completes in minutes with no interruption to your workload. Storage scale-out is only available for instances using Standard, Performance, or Capacity storage. If your instance uses local SSDs or HDDs, add nodes instead to increase storage capacity.
How long does it take to enable cold storage? What is the impact on my business during the process?
Enabling cold storage restarts each node in sequence. Each node takes about 10 to 15 minutes, so a four-node instance takes about 60 minutes. Nodes with more regions take longer.
The instance continues handling reads and writes during the process, but you may see temporary latency spikes and brief connection drops. Make sure the client is configured to reconnect automatically.
How long does it take to scale out the cold storage capacity of an instance? What is the impact on my business during the scale-out?
Cold storage scale-out completes in minutes with no interruption to your workload.
How long does it take to upgrade the configurations of an instance? What is the impact on my business during the upgrade?
Configuration upgrades apply to each node in sequence. Each node takes about 10 to 15 minutes, so a four-node instance takes about 60 minutes. Nodes with more regions take longer.
The instance continues handling reads and writes during the upgrade, but you may see temporary latency spikes and brief connection drops. Make sure the client is configured to reconnect automatically.
Scale-in and configuration downgrade
How long does it take to scale in the nodes of an instance?
Use the following formula to estimate the data migration time:
Migration time = Data to migrate ÷ (Total nodes × 30 MB/s)Example: A 10-node instance with 10 TB of used storage needs to remove 2 nodes. Data is distributed evenly, so 2 TB must migrate.
2 TB ÷ (10 nodes × 30 MB/s) = 6,990.5 seconds ≈ 2 hoursAdd a 0.5-hour cooldown period, for a total of about 2.5 hours.
This is an estimate. Actual time varies based on workload and storage medium.
Before you submit a scale-in request, verify that the remaining nodes have enough disk space to absorb the data being migrated. If any storage medium would exceed 85% utilization after scale-in, the request cannot be submitted.
For refund rules that apply after scale-in, see Unsubscription rules for Alibaba Cloud International Website.
What is the impact on my business during the node scale-in?
The instance continues handling reads and writes throughout the scale-in with no interruption.
How long does it take to downgrade the configurations of an instance? What is the impact on my business during the downgrade?
Configuration downgrades apply to each node in sequence, restarting each node in turn. Each node takes about 10 to 15 minutes, so a four-node instance takes about 60 minutes. Nodes with more regions take longer.
The instance continues handling reads and writes during the downgrade, but you may see temporary latency spikes and brief connection drops. Make sure the client is configured to reconnect automatically.
For refund rules that apply after configuration downgrade, see Unsubscription rules for Alibaba Cloud International Website.
Minor version upgrade and instance restart
How long does it take to upgrade the minor version of an instance?
Each node takes about 10 to 15 minutes. Upgrades are applied sequentially, so a four-node instance takes about 60 minutes—for both LindormDFS and LindormTable.
Can I select a minor version for the upgrade?
No. The minor version is always upgraded to the latest stable, forward-compatible version. You cannot specify a target version.
What is the impact on my business during the minor version upgrade?
-
LindormDFS: Running workloads are basically not affected.
-
LindormTable: The instance continues handling reads and writes, but you may see temporary latency spikes and brief connection drops. Make sure the client is configured to reconnect automatically.
How long does it take to restart an instance? What is the impact on my business during the restart?
Nodes restart in sequence. Each node takes about 10 to 15 minutes, so a four-node instance takes about 60 minutes. Nodes with more regions take longer.
The instance continues handling reads and writes during the restart, but you may see temporary latency spikes and brief connection drops. Make sure the client is configured to reconnect automatically.
Storage space
Understanding storage usage
Storage usage is reported differently depending on the storage type.
-
Cloud storage (Performance, Standard, and Capacity): usage includes two parts. For example, if only LindormTable is enabled, the console shows:
Label Description ① Space used by a single replica. The replica count is not factored in. ② Space used by non-engine components: recycle bin data, block metadata, pre-allocated space for blocks being written, system files, and reserved space for the distributed system. This is typically a very small proportion of total usage. 
-
Non-cloud storage (local SSD or HDD): the replica count is factored in. For example, with 3 replicas, writing 100 MB of data consumes 300 MB of storage space.
