An overview of Nomad; what it does and how it can benefit your organization. An introduction to the features of Nomad is provided, alongside a high-level explanation of the Nomad architecture.

Why use Nomad?

Conventional methods of software deployment are costly, complex and manual. With Nomad, you can achieve huge savings across your Configuration Manager infrastructure while actively improving your business’ performance. Its bandwidth throttling removes the competition between IT and business traffic and its peer-to-peer delivery eliminates servers and complexity from the network. You can treat complex, distributed networks as simple and local, delivering savings across the enterprise.

Nomad provides an efficient file transfer mechanism that integrates with Configuration Manager, providing a host of features that maximize download efficiency and minimize WAN link bandwidth usage. Advanced features, like Nomad pre-caching, enable Nomad to ease the network impact of large file distributions, such as OS deployment, with minimal effect on network users. Using Nomad together with Configuration Manager makes it possible to reduce the number of Configuration Manager servers deployed and is particularly useful in branch office scenarios where reducing WAN traffic is a critical consideration.

Nomad consists of two main components:

  • Nomad - content transfer utility enabling delivery of content efficiently, safely and securely
  • PXE Everywhere 3.2 - supporting Configuration Manager OS Deployment (OSD) to bare metal machines residing on a branch

A brief overview

Nomad provides an efficient file transfer mechanism that integrates with Configuration Manager. Its support for multicast and local re-distribution of data enables the reduction of dedicated Configuration Manager servers, particularly in branch office scenarios where reducing WAN traffic is a critical consideration.

Nomad's features provide solutions to scenarios that may arise in a core and branch network environment where Configuration Manager has been implemented, and where the branch networks have few or no servers. The typical scenarios are described in the following headings.

On this page:

Remote clients on a branch network

In this scenario, the Configuration Manager distribution point is on the core network and the remote clients are on the branch network. The core network is more liable to have servers provisioned for the purpose of distributing packages – the downside in this scenario is that the clients retrieve the packages individually.

If more than one client requires the same package it is transferred across the WAN multiple times, resulting in an inefficient use of bandwidth – there is no awareness of the end-to-end bandwidth availability and is liable to cause WAN saturation. When using Configuration Manager and BITS, the transfer rate can be capped but this is not dynamic and so cannot take advantage of quiet periods to speed up transfers.

Remote clients on the branch network

Remote distribution point on a branch network

In this scenario, the branch network has its own provisioned server acting as the Configuration Manager distribution point. The packages required at the branch are downloaded only once, thereby efficiently using the WAN, but the servers are expensive to commission and maintain. The cost of this scenario in typical branch environments can be prohibitive.

Remote desktop DP on branch network

Remote desktop distribution point on a branch network

In this scenario, desktop computers at the branch are configured to act as distribution points. The desktops, which are inherently more vulnerable than servers, must be designated and managed as if they were servers and must be constantly available. Microsoft recommends that at least two desktops are configured per site for resilience. In addition, the Microsoft BranchCache implementation of this may initially result in the packages being downloaded more than once until a clear leader becomes apparent.

Remote DP on branch network

OS deployment on a branch network

In this scenario, operating system deployments are being pushed out to machines on the branch network. The bare metal machines at the branch require a PXE service point to provide the bootstrap OS before it is applied and the PXE service point can only be implemented on a server. In addition, client machines require storage for user state, so a state migration point with local storage is required as well.

OS deployments on branch network

Now that you have an understanding of the common Configuration Manager scenarios for branch network download distribution, take a look at Nomad features which describes a number of key Nomad features that assist in the distribution of software to branch networks over WAN links.