Read this before you purchase and use Alibaba Cloud Edge Node Service (ENS).
Key limits at a glance
| Resource | Limit | Adjustable |
|---|---|---|
| Service availability (SLA) | 99.9% | No |
| Network cutover per node | Once per month, up to 6 hours | No |
| Per-tenant per-node bandwidth | 5 Gbps | Yes — contact sales or submit a ticket |
| Node retirement notice | At least 15 days in advance | No |
Node network
The average ping packet loss rate from a single edge node to a monitoring node in the same province or an adjacent province under the same carrier is no more than 1%. Special terms apply for Xinjiang and Tibet.
Node resources
Disk types. Edge nodes support local disks and ultra disks. Available disk types vary by node.
Local disk reliability. A local disk resides on a single host and inherits that host's reliability. Local disks are subject to single point of failure and data loss. If your application requires high data reliability, implement data redundancy at the application layer.
Resource usage rights. ENS grants you full usage rights to the node resources provisioned under your subscription — and only those resources. Using resources outside your subscription without authorization may result in ENS taking action against those unauthorized resources and their associated instances.
Bandwidth sharing. All tenants on a single edge node share the same public bandwidth egress, billed by usage. Per-tenant, per-node bandwidth is capped at 5 Gbps. To request additional bandwidth, contact the sales team or submit a ticket.
Node O&M
Service availability
ENS guarantees at least 99.9% service availability as defined in the Service Level Agreement. This SLA excludes:
Carrier core network failures
Scheduled maintenance announced in advance by Alibaba Cloud
Network or device failures outside Alibaba Cloud's control
Issues caused by customers
Network maintenance
ENS performs network cutovers on edge nodes no more than once per month per node. Each cutover lasts no longer than six hours.
Carrier networks may experience unexpected outages or jitter. ENS monitors node networks 24/7 and reports events to Cloud Monitor. Configure Cloud Monitor to send alerts via email, phone, or API to stay informed about node network status in real time.
Node retirement
When an edge node is scheduled for retirement due to supplier changes or other factors, ENS notifies you at least 15 days in advance. Replacement nodes — within the same province and under the same carrier — are listed on Cloud Monitor. Configure email or phone alerts in Cloud Monitor to receive retirement notifications.
On the retirement date, the node stops service and releases all resources. Before that date:
Back up your data.
Release your resources on the node.
Migrate your workloads to the replacement node.
Failure scenarios
Host failure (local disk instances). If a host running a local disk instance fails, ENS migrates the instance to another host within the same node to restore connectivity as soon as possible. Data reliability is not guaranteed — back up critical data at the application layer.
Edge node unreachable. If an entire edge node becomes unreachable, ENS migrates your instance to another node in the same region to restore regional connectivity as soon as possible. The instance ID remains unchanged, but the instance IP address changes. Data reliability is not guaranteed.