NLINK deployment for high availability

Owned by Rakesh
Last updated: Mar 02, 2016 by Junot Systems
2 min read

This topic provides details of some options of how NLINK can best be deployed in a 24x7 operation.

Unplanned downtime, as opposed to planned downtime, is a random, unpredictable event. The systems can be unavailable because of hardware issues or in some cases software issues. Depending on the overall business process, and role NLINK plays in such setup, there are various options available to ensure high availability of the NLINK Server.

Typically, the NLINK Server interacts with multiple systems like SAP, SQL Server databases or file systems. This article does not cover ensuring availability of all such systems. Individual system requirements vary greatly and need to be addressed on a case-by-case basis.

 

Clustering

A failover clustering environment can be used to ensure high availability for planned or unplanned downtime. Planned downtime incldues cases where various software upgrades need to occur and unplanned downtime incldues cases that are caused by hardware or software failures.

NLINK can be deployed seamlessly on any Microsoft Windows based clustering environment. The clustering environment can be software based (Microsoft Cluster Server, Stratus everRun, etc.) or at the hardware level (for example, Stratus ftServer Systems).  Additionally, the cluster setup can be based on virtual environments, such as VMware or Microsoft’s Hyper-V.  None of these cluster based deployments require any changes to the NLINK product or the NLINK meta-database. At a technical level, all nodes in the cluster need to be configured to be running the same version of NLINK and using the same NLINK meta-database 

The advantage of a clustered deployment is that all nodes appear identical to other systems. So there will not be any difference for either ADC (Automated Data Collection) users or 3rd party software systems connecting to NLINK.

NLINK has been deployed many times in a failover clustering environment. In fact, clustering is a standard deployment scenario routinely used by some of Junot’s embedded partners.

Non-Clustering Options

In some scenarios, the cost of clustering may not justifiable. In such cases there are some low cost approaches that can be considered to improve availability short of deploying on a cluster server. 

  1. If the NLINK Server is hosted in an environment where network link failures are the main issue, multi-homing can be used. In such a setup the host connects to two different subnets using dual network cards. As long as the network landscape allows access to all required systems on both subnets, such an option can be used.  Most computer servers offer the option of having multi-homed hosts at a very low cost. Of course, the best option would be to have the actual network subnet issue solved, but where this is not possible multi-homed hosts can help improve availability.
  2. If the NLINK Server is hosted in a virtual environment, then any downtime can be minimized by ensuring that there is an identical instance of a standby host with NLINK setup accordingly. Care must be taken to ensure that both systems are identical, have the same version of the NLINK Server installed and are configured to run the same copy of NLINK Meta-database. In addition, both the instances of the virtual systems should be hosted on different physical hardware to avoid a single point of failure,

    Neither of the above two options will guarantee 100% up-time but they can provide improved availability if care is taken to monitor and switch the systems in case of issues.