It’s common for a newly released application to have a few bugs in it. Customers may grumble, and IT service requests may increase, but life goes on and people will figure out how to work around the issues. The same cannot be said for a network software upgrade. If the network goes down, everything grinds to a halt, and service to employees and customers ceases.
In the old hardware network model, an operator relied on three to four vendors to run annual or semiannual upgrades, and even that process put the network’s resiliency at risk. As networks transition to a software model, they are supported by a myriad of best-of-breed partners and a more diverse ecosystem, increasing the frequency of updates and therefore the degree of network vulnerability. In addition, the technology implemented is less mature, resulting in greater network exposure to errors and risk.
A network going down is extremely costly not only because of the cessation of connectivity and work; the reputational cost to companies can be long lasting and sometimes irreversible. The traditional upgrade processes simply will not scale in the new world of software defined networks (SDN) and virtual network functionality (VNF).
To avoid this risk, network management must develop software engineering lifecycle management, the type of discipline found in software IT devops. However, because of the critical nature of network resiliency combined with the increase in complexity and decrease in technological maturity, the processes must be more detailed, more precise, more coordinated and more test-heavy than traditional devops. It needs to be devops on steroids.
Network devops require a system that embraces constant, asynchronous updates and rigorous testing of new and often less mature technologies. Unlike software IT devops, there are no off-the-shelf platforms for network devops management. However, there are platforms being developed that operators can customize to help automate these processes. The resiliency of the network is paramount so rigorous testing is a core requirement, and testing will be even more demanding than before because of the relative immaturity of the technology. This would involve rethinking the process used for continuous testing/development/integration (CT/CD/CI) in the IT world and require customizing the standard tool kit available for DevOps today.
Typically, one would need to test the network software patches every six weeks, testing for not just functional integrity but also stress testing, interoperability and feature set compatibility testing. At the same time, there will be software extensions developed by the service providers that would need to be tested for consistency. This level of complexity and maturity requires a very comprehensive DevOps environment for Network development, engineering and operations
Not only are the existing processes inadequate for these challenges, but those tasked with maintaining the network are unlikely to have the agile devops skills needed. CTIOs should execute a complete skills assessment and a comprehensive set of training workshops to meet these new demands. Even with training, new talent with agile devops expertise will be needed. Many of these developers are in high demand, so the recruitment and retention of this talent will be an ongoing and demanding task for the organization.
The coordination of multi-speed IT is also required to transition to a digital network. This is a complex challenge both in terms of skill, technology, culture and coordination. The existing network team tasked with implementing and managing upgrades to the hardware components are essential to the ongoing operation and success of your enterprise.
At the same time, a company must introduce a new set of employees with different skills and backgrounds to work closely together and assure all updates are well-tested and seamlessly implemented with a high degree of confidence. This will require transparency and very close communication and collaboration. The two groups are likely to operate quite differently, come from different professional backgrounds and possess different skill sets, making a culture of respect and appreciation for what each group brings to the success of the organization essential.
Management and leadership
The challenges ahead are formidable, and management must be empowered from the C-suite down to drive this transformation. This is a highly complex, interconnected process fraught with risks to lose sight of the goal and for one initiative to get out of synch with another. Mistakes are to be expected and rebalancing will be ongoing. Success will require more than management; it requires leadership, because when things go wrong, as they will, leaders will need to have the confidence of their teams for teams to stay on track, motivated and focused on the end goal.
The key is to have a plan and communicate regularly and effectively with frequent, in-depth checkpoints. There are playbooks on how to make this journey, but each organization’s challenges, assets and goals will be unique.
The art of balance
The key to success is balancing this process across all the factors we have identified and to understand the challenges an organization faces:
- How rapidly can you and should you undergo this transition?
- Are you adequately prepared to initiate the transformation – planning, capital and talent?
- Will the core functionality of your network remain resilient throughout this process?
- Will your network be ready to capitalize on your business needs and opportunities?
- Have you solicited the help of experts you need to advise you on your journey?
The journey of developing a pervasive network that reduces operating costs while enabling the competitive advantages of a modern digital enterprise starts with the vision and commitment of management. The above challenges are far from trivial, but if management has a well-thought-out plan, the self-awareness to identify what they must learn and the patience and assets to see it through, it is not only possible, it is essential to competing in the software driven industry.
This article is published as part of the IDG Contributor Network. Want to Join?
<< This article was originally published on blog here. >>