You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: content/cumulus-netq-51/Installation-Management/Install-NetQ/Before-You-Install.md
+7-11Lines changed: 7 additions & 11 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -9,7 +9,7 @@ This overview is designed to help you understand the various NetQ deployment and
9
9
10
10
## Installation Overview
11
11
12
-
Consider the following deployment options and requirements before you install the NetQ system:
12
+
Consider the following deployment options and requirements before you install the NetQ system.
13
13
14
14
| Single Server | Cluster| Scale Cluster |
15
15
| --- | --- | --- |
@@ -22,29 +22,25 @@ Consider the following deployment options and requirements before you install th
22
22
23
23
*When switches are {{<linktitle="Integrate NetQ with Grafana/#requirements-and-support"text="configured with both OpenTelemetry (OTLP)">}} and the NetQ agent, switch support per deployment model is reduced by half.
24
24
25
-
## Deployment Type: On-Premises or Cloud
26
25
27
-
**On-premises deployments** are hosted at your location and require the in-house skill set to install, configure, back up, and maintain NetQ. This model is a good choice if you want very limited or no access to the internet from switches and hosts in your network.
28
-
29
-
In the **cloud deployment**, you host only a small, local server on your premises that connects to the NetQ cloud service over selected ports or through a proxy server. NetQ cloud supports local data aggregation and forwarding---the majority of the NetQ applications use a hosted deployment strategy, storing data in the cloud. NVIDIA handles the backups and maintenance of the application and storage.
26
+
## Server Arrangement: Single or Cluster
30
27
31
28
In all deployment models, NetQ agents reside on the switches and hosts they monitor in your network.
32
29
33
-
## Server Arrangement: Single or Cluster
34
-
35
30
### Single Server
36
31
37
32
A standalone server is easier to set up, configure, and manage, but limits your ability to scale your network monitoring. Deploying multiple servers allows you to limit potential downtime and increase availability by having more than one server that can run the software and store the data. Select the standalone, single-server arrangement for smaller, simpler deployments.
38
33
39
34
### Cluster of Servers
40
35
41
-
NVIDIA offers two types of cluster deployments: cluster and scale cluster. Both deployments are available on-premises and offer high-availability and provide redundancy in case of node failure.
36
+
NVIDIA offers two types of cluster deployments: cluster and scale cluster. Both deployments are available on-premises and offer high-availability to provide redundancy in case of node failure.
37
+
38
+
The **cluster** implementation comprises three servers: one master and two workers nodes. NetQ supports high availability using a virtual IP address. Even if the master node fails, NetQ services remain operational. This deployment supports networks with up to 100 switches.
42
39
43
-
The **cluster**implementation comprises three servers: one master and two workers nodes. NetQ supports high availability using a virtual IP address. Even if the master node fails, NetQ services remain operational. However, keep in mind that the master hosts the Kubernetes control plane so anything that requires connectivity with the Kubernetes cluster—such as upgrading NetQ or rescheduling pods to other workers if a worker goes down—will not work. During the installation process, you configure a virtual IP address that enables redundancy for the Kubernetes control plane. In this configuration, the majority of nodes must be operational for NetQ to function. For example, a three-node cluster can tolerate a one-node failure, but not a two-node failure. For more information, refer to the {{<exlinkurl="https://etcd.io/docs/v3.3/faq/"text="etcd documentation">}}.
40
+
The **scale cluster**deployment supports large networks and allows you to adjust NetQ's network monitoring capacity by adding additional nodes to your cluster as your network expands. For example, you can deploy a three-node scale cluster that accommodates up to 1,000 switches. When you add switches to your network, the extensible framework allows you to add additional nodes to support a greater number of switches. NVIDIA recommends this option for networks comprising 100 or more switches with 100 or more interfaces per switch.
44
41
45
-
As an alternative to the three-node cluster deployment with a virtual IP address, you can use an external load balancer to provide high availability for the NetQ API and the NetQ UI.
42
+
In both cluster deployments, the majority of nodes must be operational for NetQ to function. For example, a three-node cluster can tolerate a one-node failure, but not a two-node failure. Similarly, a 5-node cluster can tolerate a two-node failure, but not a three-node failure. If the majority of failed nodes are Kubernetes control plane nodes, NetQ will no longer function. For more information, refer to the {{<exlinkurl="https://etcd.io/docs/v3.3/faq/"text="etcd documentation">}}.
46
43
47
-
The **scale cluster** deployment supports large networks and offers the highest level of scalability, allowing you to adjust NetQ's network monitoring capacity by adding additional nodes to your cluster as your network expands. For example, you can deploy a three-node scale cluster that accommodates up to 1,000 switches. When you add more switches to your network, the extensible framework allows you to add two additional nodes that can accommodate a network of up to 2,000 switches. NVIDIA recommends this on-premises option for networks comprising 100 or more switches with 100 or more interfaces per switch.
48
44
49
45
{{%notice note%}}
50
46
Large networks have the potential to generate a large amount of data. For large networks, NVIDIA does not recommend using the NetQ CLI; additionally, {{<linktitle="Access Data with Cards/#table-settings"text="tabular data in the UI">}} is limited to 10,000 rows. If you need to review a large amount of data, NVIDIA recommends downloading and exporting the tabular data as a CSV or JSON file and analyzing it in a spreadsheet program.
0 commit comments