1---
2layout: "docs"
3page_title: "Performance Standby Nodes - Vault Enterprise"
4sidebar_title: "Performance Standbys"
5sidebar_current: "docs-vault-enterprise-perf-standbys"
6description: |-
7  Performance Standby Nodes - Vault Enterprise
8---
9
10# Performance Standby Nodes
11
12Vault supports a multi-server mode for high availability. This mode protects
13against outages by running multiple Vault servers. High availability mode
14is automatically enabled when using a data store that supports it. You can
15learn more about HA mode on the [Concepts](/docs/concepts/ha.html) page.
16
17Vault Enterprise offers additional features that allow HA nodes to service
18read-only requests on the local standby node. Read-only requests are requests
19that do not modify Vault's storage.
20
21## Server-to-Server Communication
22
23Performance Standbys require the request forwarding method described in the [HA
24Server-to-Server](/docs/concepts/ha.html#server-to-server-communication) docs.
25
26A performance standby will connect to the active node over the existing request
27forwarding connection. If selected by the active node to be promoted to a
28performance standby it will be handed a newly-generated private key and certificate
29for use in creating a new mutually-authenticated TLS connection to the cluster
30port. This connection will be used to send updates from the active node to the
31standby.
32
33## Request Forwarding
34
35A Performance Standby will attempt to process requests that come in. If a
36storage write is detected the standby will forward the request over the cluster
37port connection to the active node. If the request is read-only the Performance
38Standby will handle the requests locally.
39
40Sending requests to Performance Standbys that result in forwarded writes will be
41slightly slower than going directly to the active node. A client that has
42advanced knowledge of the behavior of the call can choose to point the request
43to the appropriate node.
44
45### Direct Access
46
47A Performance Standby will tag itself as such in consul if service registration
48is enabled. To access the set of Performance Standbys the `performance-standby`
49tag can be used. For example to send requests to only the performance standbys
50`https://performance-standby.vault.dc1.consul` could be used (host name may vary
51based on consul configuration).
52
53### Behind Load Balancers
54
55Additionally, if you wish to point your load balancers at performance standby
56nodes, the `sys/health` endpoint can be used to determine if a node is a
57performance standby. See the [sys/health API](/api/system/health.html) docs for
58more info.
59
60## Disabling Performance Standbys
61
62To disable performance standbys the `disable_performance_standby` flag should be
63set to true in the Vault config file. This will both tell a standby not to
64attempt to enable performance mode and an active node to not allow any
65performance standby connections.
66
67This setting should be synced across all nodes in the cluster.
68
69## Monitoring Performance Standbys
70
71To verify your node is a performance standby the `vault status` command can be
72used:
73
74```
75$ vault status
76Key                                    Value
77---                                    -----
78Seal Type                              shamir
79Sealed                                 false
80Total Shares                           1
81Threshold                              1
82Version                                0.11.0+prem
83Cluster Name                           vault-cluster-d040e74c
84Cluster ID                             9f82e03b-71fb-97a6-9c5a-46fa6715d6e4
85HA Enabled                             true
86HA Cluster                             https://127.0.0.1:8201
87HA Mode                                standby
88Active Node Address                    http://127.0.0.1:8200
89Performance Standby Node               true
90Performance Standby Last Remote WAL    380329
91```
92