Plan Your Deployment

This chapter contains the following topics:

Before You Begin

This section explains the decisions you need to make before installing Crosswork Network Controller solution on your preferred platform.

  1. Determine how many VMs you need

  2. Identify the resource footprint

  3. Special Considerations

After completing the planning in the above steps, follow the relevant installation workflow steps for your platform:


Note

Starting with the Crosswork Network Controller version 4.1, Crosswork deployment is no longer supported for the Cisco CSP platform. For more information, see End-of-Life Announcement for the Cisco Cloud Services Platform Operating System.

Determine how many VMs you need


Attention

This topic is applicable only when you need to deploy Crosswork Network Controller as a cluster. In case of single VM deployment, see Install Cisco Crosswork Network Controller on a Single VM

After finalizing the Crosswork Network Controller tier that meets the needs of your production environment, use the table below to determine the number of VMs required to deploy the Crosswork Network Controller cluster and the type of Data Gateways to be deployed. This is a crucial step as each Crosswork Network Controller tier can impact the overall resources needed.


Note

Geo redundancy solution requires double the number of VMs. For more information, see Enable Geo Redundancy section.

Crosswork Network Controller is available in these tiers:

Table 1. Crosswork Network Controller tiers

Cisco Crosswork Network Controller tier

1

Contents

Crosswork Data Gateway deployment

2

Minimum number of VMs needed

3

Essentials

Element Management Functions

On-Premise Standard (default): Collectors only.

3 Hybrid nodes

Path Analytics

 On-Premise Standard

Advantage

Crosswork Optimization Engine

On-Premise Standard (default): Collectors only.

3 Hybrid nodes + 2 Worker nodes

Crosswork Active Topology

On-Premise Standard (default): Collectors only.

Crosswork Service Health

On-Premise Extended: Collectors and offload services.

Premier

Crosswork Workflow Manager

On-Premise Extended: Collectors and offload services.

3 Hybrid nodes + 2 Worker nodes

Crosswork Workflow Manager Solutions

On-Premise Extended: Collectors and offload services.

Add-on

Crosswork Change Automation

On-Premise Extended: Collectors and offload services.

3 Hybrid nodes + 2 Worker nodes

Crosswork Health Insights

On-Premise Extended: Collectors and offload services.

1

There are licensing implications for different tiers, please consult your Cisco Account team to understand which tiers and licenses are required for your use cases.

2

The VM resource requirements for Crosswork Data Gateway are different for each type and cannot be modified. Therefore, if your requirements change, you must re-deploy the Crosswork Data Gateway to move from one type to another. For more information, see the Redeploy a Crosswork Data Gateway VM section in the Cisco Crosswork Network Controller 7.2 Administration Guide.

3

The number of VMs mentioned is the minimum count. You can add more Worker nodes (maximum up to three worker nodes) as needed. If your requirements exceed the minimum count, please contact the Cisco Customer Experience team.

Ensure that you have sufficient worker nodes in your cluster. You can always check the load in your cluster and choose to add new worker nodes post installation. For more information, see the Deploy New Cluster Nodes section in the Cisco Crosswork Network Controller 7.2 Administration Guide.

Identify the resource footprint

Once you determine the Crosswork applications you want and the number of VMs you will need to deploy to host them, ensure that you have the resources needed for them. The resources required per VM such as CPU, Memory, and Storage vary based on the data center where your VMs will be hosted.


Note

  • The resources listed for NSO are higher than for other NSO use cases due to the additional requirements Crosswork Network Controller places on NSO.

  • The NSO footprint depends on the type of deployment, standalone or LSA.

  • The SR-PCE count will depend on the number of head-ends that need to be managed

  • The values in Storage column is the space needed for storing Crosswork files and does not consider any additional overhead that may be required (for example, RAID configuration). Storage requirements vary based on factors such as the number of devices being supported and the type of deployment selected. However, 1 TB disk space should work for most deployments.

  • The storage required for each backup will vary based on the your cluster size, applications in the cluster, and the scale requirements.

  • Solid-state drives (SSDs) are required for all Crosswork Network Controller deployments due to their performance.

  • Upgrade of the cluster (using parallel hardware) temporarily requires double the total disk space used by the cluster. See Upgrade Using Parallel Hardware for more information.

  • The number of Data Gateways needed depends on the number of devices you have in your network and the level of redundancy you want (1:n up to 1:1). To determine the number of Crosswork Data Gateways needed, contact the Cisco Customer Experience team.

See these topics to know more about resource requirements per VM for each deployment:

VM requirements

This topic explains the requirements for Hybrid and Worker VMs when deploying different tiers of the Crosswork Network Controller.

VM storage specifications

Table 2. Storage specifications per VM

Crosswork Network Controller tier

VM type

Device count

Recommended VM disk size (approximated)

File system to configure

Required file system size

Essentials

Hybrid

<= 15k

~ 1 TB

ManagerDataFs

500 GB

Hybrid

> 15k

~ 2 TB

ManagerDataFs

1500 GB

Worker

<= 15k

~ 1 TB

WorkerDataFs

500 GB

Worker

> 15k

~ 2 TB

WorkerDataFs

1500 GB

Advantage / Premier

Hybrid

N/A

~ 1 TB

ManagerDataFs

500 GB

Worker

<= 15k

~ 1 TB

WorkerDataFs

500 GB

Worker

> 15k

~ 2 TB

WorkerDataFs

1500 GB

Here are the disk parameter values for each use case:


Note

It is recommended to use the default value unless otherwise specified.
Table 3. Disk parameter values

Disk parameter

1 TB VM disk storage

2 TB VM disk storage

Datafs disk size (GB)

485 GB (default) – 8000 GB

1500 GB (default) – 8000 GB

Logfs disk size (GB)

20 GB (default) – 1000 GB

Corefs partition size (GB)

18 GB (default) – 1000 GB

High speed disk size (GB)

15 GB (default)

Minimum backup partition size (percentage of the data disk space)

1% – 80%

Default value: 35%
Table 4. Data file systems

Parameter

Description

ManagerDataFsSize

Size of data file system for Hybrid VMs.

WorkerDataFsSize

Size of data file system for Worker VMs.

Disk requirements

The VM data store(s) need to have disk access latency < 10 ms and > 4000 IOPs. For more information on the diagnostic checks, see Diagnostic assessment.

Table 5. IOPs requirements

IOPs value

Status

Response

IOPs value > 4000

 PASS

Installation proceeds without the need for user acknowledgement.

1000 < IOPs value > 4000

 NOT IDEAL

For production deployments, the installation requires ~ 4000 IOPs for optimal scaling performance. If the system operates in the suboptimal range of 1000 to 4000 IOPs, the installation process requires user acknowledgment before continuing.

If the installation is blocked, you can choose to ignore the failure and continue. You must explicitly acknowledge the report before proceeding, and by doing so, you acknowledge this warning and accept the risks.

IOPs value < 1000

 FAIL

Installation is blocked irrespective of user response.

Important

  • When installing on shared network storage or datastores used by multiple Crosswork virtual machines (VMs), power on Crosswork nodes sequentially. Start with the seed node first, then power on each additional node one at a time. This sequential approach ensures accurate IOPs diagnostics across all Crosswork nodes.

  • If you power on nodes simultaneously in the shared network storage or datastore environments, initial diagnostics might not accurately reflect IOPs availability, which can lead to incorrect system assessments.

  • The interval time between powering on VMs during the initial installation depends on how long diagnostics take to complete. The diagnostic all command typically completes in 4–5 minutes in VMware environments and in 14–15 minutes in KVM environments.

Table 6. Network bandwidth requirements

Network bandwidth

Status

Response

Bandwidth > 8000 Mbps

PASS

Installation proceeds without the need for user acknowledgement.

1000 < Bandwidth < 8000 Mbps

NOT IDEAL

Values between 1000 and 8000 Mbps are considered suboptimal. Installation is blocked, but you can choose to ignore the failure and continue. You must explicitly acknowledge the report before proceeding, and by doing so, you acknowledge this warning and accept the risks.

Bandwidth < 1000 Mbps

FAIL

Installation is blocked irrespective of user response.

Resource footprint for VMware

This topic explains the resource requirements per VM to deploy Crosswork Network Controller Hybrid or Worker node, Crosswork Data Gateways, NSO, and SR-PCE on VMware.


Important

  • As Crosswork Network Controller cluster nodes place high demands on the VMs, ensure that you have not oversubscribed CPU or memory resources on the machines hosting the nodes.

  • Ensure that you have a docker-capable host to load the installer tool.

  • The VM data store(s) must have disk access latency < 10 ms and > 4000 IOPS.

  • The network bandwidth from Node IP towards to Management VIP should be at least 8000 Mbps.

Table 7. Resource footprint for VMware

Component

vCPU

Clock Freq (GHz)

Memory (RAM)

Network Interface Controller (NIC)

Storage (Boot disk + Data disks)

Crosswork Network Controller Hybrid or Worker node

12

Minimum clock reservation: 18 GHz

>= 2.20

96 GB

10 Gbps

 See VM requirements

Crosswork Data Gateway On-Premise Standard

12

>= 2.20

48 GB

10 Gbps

70 GB (50 GB + 20 GB)

Crosswork Data Gateway On-Premise Extended

20

>= 2.20

96 GB

10 Gbps

570 GB (50 GB + 520 GB)

Cisco NSO

16

>= 2.20

256 GB

10 Gbps

1 TB

Cisco SR-PCE

Note

 

This is the requirement for running a Cisco XRv9K with SR-PCE functionality enabled.

8

>= 2.20

24 GB

10 Gbps

70 GB

Basic SCP Server (for storing backups)

-

-

-

-

At least 25 GB (recommended)

Resource footprint for AWS EC2

This topic explains the resource requirements per VM to deploy Crosswork Network Controller Hybrid or Worker node, Crosswork Data Gateways, NSO, and SR-PCE on AWS EC2.

When deploying in AWS EC2, choose an EC2 instance size that meets or surpasses these resource requirements listed:


Important

The VM data store(s) must have disk access latency < 10 ms and > 4000 IOPS.
Table 8. Resource footprint for AWS EC2

Component

vCPU

Memory (RAM)

Network Interface Controller (NIC)

Storage (Boot disk + Data Disks)

Crosswork Network Controller Hybrid or Worker node

12

96 GB

10 Gbps

 See VM requirements

Crosswork Data Gateway On-Premise Standard

12

64 GB

10 Gbps

70 GB (50 GB + 20GB)

Crosswork Data Gateway On-Premise Extended

24

128 GB

10 Gbps

570 GB (50 GB + 520 GB)

Cisco NSO

16

256 GB

10 Gbps

1 TB

Cisco SR-PCE

Note

 

This is the requirement for running a Cisco XRv9K with SR-PCE functionality enabled.

8

24 GB

10 Gbps

120 GB

Basic SCP Server (for storing backups)

-

-

-

At least 25 GB (recommended)

Resource footprint for KVM

This topic explains the resource requirements per VM to deploy Crosswork Network Controller Hybrid or Worker nodes, Crosswork Data Gateways, and NSO on KVM based hypervisor.


Important

  • As Crosswork Network Controller cluster nodes place high demands on the VMs, ensure that you have not oversubscribed CPU or memory resources on the machines hosting the nodes.

  • The VM data store(s) must have disk access latency < 10 ms and > 4000 IOPS.

  • The network bandwidth from Node IP towards to Management VIP should be at least 8000 Mbps.

Table 9. Resource footprint for KVM

Component

vCPU

Clock Freq (GHz)

Memory (RAM)

Network Interface Controller (NIC)

Storage (Boot disk + Data disks)

Crosswork Network Controller Hybrid or Worker node

12

Minimum clock reservation: 18 GHz

>= 2.20

96 GB

10 Gbps

 See VM requirements

Crosswork Data Gateway On-Premise Standard

12

>= 2.20

48 GB

10 Gbps

70 GB (50 GB + 20 GB)

Crosswork Data Gateway On-Premise Extended

20

>= 2.20

96 GB

10 Gbps

570 GB (50 GB + 520 GB)

Cisco NSO

16

>= 2.20

256 GB

10 Gbps

1 TB

Basic SCP Server (for storing backups)

-

-

-

-

At least 25 GB (recommended)

Resource footprint for single VM deployments

This table lists the recommended resources to deploy Crosswork Network Controller on a single VM.


Important

  • As Crosswork Network Controller node places high demands on the VM, ensure that you have not oversubscribed CPU or memory resources on the machine hosting the node.

  • The VM data store need to have disk access latency < 10 ms and > 4000 IOPS.

  • The network bandwidth from Node IP towards to Management VIP should be at least 8000 Mbps.

Table 10. Crosswork Network Controller on a single VM

VM profile

vCPU

Memory (RAM)

Storage

Use case
XLarge

24

128 GB

1 TB

The XLarge VM profile supports deployment of all tiers (Essential, Advantage, and Premier) of Crosswork Network Controller on a single VM.

Large

12

96 GB

1 TB

The Large VM profile only supports deployment of Crosswork Network Controller Essentials tier on a single VM.

Small

8

48 GB

650 GB

The Small VM profile is only used to deploy the arbiter VM on a single VM (to enable auto-arbitration in a geo HA setup).

Host resource requirements

This topic details the recommended host requirements for deploying the various tiers of the Crosswork Network Controller.

Formulas for resource calculation

When sizing your host machine, allocate a 20% buffer for CPU and memory resources, and add 100 GB of storage for each VM for each VM that will run on the host machine. This helps ensure smooth performance and reduces the risk of resource-related issues during operation.


Remember

Please note that this is a recommendation for optimal performance, not a mandatory requirement.

  • RAM calculation:

    Total RAM: (Number of VMs) * (RAM per VM)

    Buffer (for example, 20%): Total RAM * 0.20

    Estimated total RAM needed (GB) = Total RAM + Buffer

  • CPU (vCPU) calculation:

    Total vCPUs: (Number of VMs) * (vCPUs per VM)

    Buffer (for example, 20%): Total vCPUs * 0.20

    Estimated total vCPUs needed = Total vCPUs + Buffer

  • Storage calculation:

    Total Storage: (Number of VMs) * (Storage per VM)

    Buffer: additional 100 GB per VM

    Estimated total storage needed (GB) = Total Storage + Buffer

Cluster deployments

Cluster deployments support only the Large VM profile.


Important

The calculations in this table are performed for a minimum cluster size of 3 VMs and assume all 3 VMs reside on a single host with 1 TB of storage per VM. If you plan to distribute your VMs across multiple hosts (for example, 1 VM per host), you must adjust the calculated host resource requirements accordingly.

Table 11. Host resource requirements for cluster deployments

Component

Minimum requirement per host

RAM

Number of VMs: 3

RAM per VM: 96 GB (Large profile)

Total RAM: 3 * 96 = 288 GB

Buffer: 288 * 0.20 = 57.6 GB

Estimated total RAM needed: 288 + 57.6 = 345.6 GB

CPU

Number of VMs: 3

vCPUs per VM: 12 (Large profile)

Total vCPUs needed for VMs: 3 * 12 = 36

Buffer: 36 * 0.20 = 7.2

Estimated total vCPUs needed: 36 + 7.2 = 43.2

Storage

Number of VMs: 3

Storage per VM: 1000 GB

Total storage needed for VMs: 1000 * 3 = 3000 GB

Buffer: 3 * 100 = 300 GB (additional 100 GB per VM)

Estimated total storage needed: 3000 + 300 = 3300 GB

Single VM deployments

Single VM deployments support Small (for arbiter only), Large, and XLarge VM size profiles to suit different use cases. For more information, see Resource footprint for single VM deployments.

Table 12. Host resource requirements for single VM deployments

Component

Minimum requirement per host

Small VM profile

Large VM profile

XLarge VM profile

RAM

RAM per VM: 48 GB

Buffer: 48 * 0.20 = 9.6 GB

Estimated total RAM needed: 48 + 9.6 = 57.6 GB

RAM per VM: 96 GB

Buffer: 96 * 0.20 = 19.2 GB

Estimated total RAM needed: 96 + 19.2 = 115.2 GB

RAM per VM: 128 GB

Buffer: 128 * 0.20 = 25.6 GB

Estimated total RAM needed: 128 + 25.6 = 153.6 GB

CPU

vCPUs per VM: 8

Buffer: 8 * 0.20 = 1.6

Estimated total vCPUs needed: 8 + 1.6 = 9.6

vCPUs per VM: 12

Buffer: 12 * 0.20 = 2.4

Estimated total vCPUs needed: 12 + 2.4 = 14.4

vCPUs per VM: 24

Buffer: 24 * 0.20 = 4.8

Estimated total vCPUs needed: 24 + 4.8 = 28.8

Storage

Storage per VM: 650 GB

Buffer: 100 GB

Estimated total storage needed: 650 + 100 = 750 GB

Storage per VM: 1000 GB

Buffer: 100 GB

Estimated total storage needed: 1000 + 100 = 1100 GB

Storage per VM: 1000 GB

Buffer: 100 GB

Estimated total storage needed: 1000 + 100 = 1100 GB

Special Considerations

In addition to the above instructions, there may be certain setup options that you need to consider before you begin the installation.

  • Check versions of NSO and SR-PCE – Please see the Release Notes for Crosswork Network Controller 7.2.0 to know the NSO and SR-PCE versions compatible with Crosswork Network Controller.

  • Are you going to use self-signed certificates? – if yes, you need to make the certificates available. For more information on the type of certificates supported and how to manage them, see the Manage Certificates section in the Cisco Crosswork Network Controller 7.2 Administration Guide.

  • Do you plan to integrate Crosswork with external authentication servers?– Integration with TACACS+ or other external authentication servers will require you to have credentials created for the Crosswork user accounts and roles.

  • In Crosswork Data Gateway, do you want to configure a Syslog server to collect the syslog?– if yes, then provide the host name or IPv4 or IPv6 address of an external syslog server. Or, you have the option to configure the syslog server using the interactive console after the installation is complete.


    Note

    IPv6 deployments allow the use of both full-form and short-form IPv6 addresses during installation. In the backend, all IPv6 addresses are normalized to a standard format regardless of the input form.

  • In Crosswork Data Gateway, do you want to configure an Auditd server to collect the event logs?– if yes, then provide the host name or IPv4 or IPv6 address of an external auditd server. As an alternative, you have the option to configure the auditd server using the interactive console after the installation is complete.

  • Do you plan to enable the automatic execution of linked playbooks? – If yes, you must enable Playbook Job Scheduling and disable Credential Prompting for playbook execution in the Network Automation settings window. For more information, see the Enable Automatic Playbook Execution topic in the Cisco Crosswork Network Controller 7.2 Closed-Loop Network Automation guide.