Kirill Kedrinski - Fotolia
Ivan Pepelnjak, writing in IpSpace, examined whether or not IT teams should buy or build a network automation system....
Pepelnjak previously argued against buying a "reassuringly-expensive" and fully integrated network automation system. But when it comes to point tools, his reasoning is exactly the opposite. As long as the tools have a well-defined edge and the ability to interact with the "outside world" through an API, he said there is no need to reinvent the wheel when deploying network automation.
Instead, he said, it is more important to control the overall system architecture and interfaces between components, which offers companies the ability to replace individual tools with options or even a homegrown replacement if needed.
When answering the question of whether to buy or to build a network automation system, Pepelnjak suggests enterprises should look into the initial cost, cost over the lifetime of the project and the time-to-value factor. Pepelnjak added there is often a temptation to reinvent the wheel, but warned that if an open source "wheel" already exists, don't build another one. Instead engineers should focus on augmenting software or building a system anew and contributing it to the open source community.
Dig deeper into Pepelnjak's thoughts on network automation systems.
Enterprises eager to adopt AI and machine learning
Mike Leone, an analyst at Enterprise Strategy Group in Milford, Mass., sees incredible traction for AI and machine learning in enterprise IT. While AI and machine learning are top priorities for companies working toward digital transformation, he said, investment remains modest as a result of the infrastructure costs associated with these new technologies. Both fields rely heavily on different elements of the technology stack, from physical hardware supporting storage, compute and networking to software that handles compliance and other requirements. Yet enterprises still struggle to have all their networking infrastructure in sync, citing security, compliance, and to a lesser extent, big data, as the "weak links" in the chain, according to Leone. A majority of organizations rely on three different tools to develop, test, deploy and manage machine learning models, ESG said.
ESG research indicates that senior IT executives are the biggest driver behind new AI initiatives and IT is far ahead of other business units in terms of adopting AI and machine learning applications. Data gathered by ESG found operational efficiency was the most significant goal, followed by risk reduction, improved security, better customer insights and prediction capabilities. "I compare the adoption of AI and ML in enterprise IT to what flash storage has done to the storage market, but on a much larger scale," Leone said. "If you aren't aboard the AI/ML rocket ship, figure out a way to get on it -- both as a customer relying on it and as a vendor providing technology to enable it. It will change the way we think about IT, and it will change the way IT operates."
Read more of Leone's ideas about AI and machine learning.
Adopting hyper-converged infrastructure
Keith Townsend of the CTO Advisor said that converged infrastructure and hyper-converged infrastructure (HCI) both aim to simplify IT yet both typically fail in large organizations -- a problem that rarely stems from technology.
When it was pioneered by companies like Nutanix, HCI was grouped as storage, compute, management and hypervisor bundled together. HCI's marketing appeal is infrastructure simplification, reducing the multiple consoles needed by administrators to manage virtualization and storage, as well as allowing users to cut design time and integration testing.
According to Townsend, scale breaks down the effectiveness of technology. "The larger the organization, the more difficult the goal of simplifying operations via technology adoption," he said. "Transformation is about people, process and technology. HCI is only a tool. Without a change in operations, HCI can add complexity to existing services."
But Townsend warned organizations that it's necessary to "fully adopt" an HCI-fueled operations model. If they don't, they run the risk of merely adding another management stack, thus increasing rather than reducing overhead.
Explore more of Townsend's assessment of HCI.