So how do network startups fit in? Let’s consider what trades off between simplicity and reduced risk. It’s the cost.
If new network equipment could save you a lot of money, improve productivity, or support a new application or service that significantly lowers overall costs, it’s worth looking into. Of the 321 enterprise IT professionals who provided opinions on network startups, 303 said cost was the only reason they rushed to join a startup, with the remaining respondents citing cost as the strongest reason.
Startups and the venture capitalists that support them are not stupid either, so we can already see through several examples how network startups solve the problem of focusing on money. There are two paths: dramatically lower costs through innovative design or focus on the network impact of new computing strategies. A good example of both can be seen in software-defined wide area networks (SD-WAN).
Enhancing security from SD-WAN providers
When SD-WAN first appeared, only network service costs were considered. Corporate networks used specialized technologies (digital trunks, carrier Ethernet) for access connections and MPLS VPNs for WANs, while user-centric Internet connections provided twice the capacity at an average of one-eighth the cost (according to the companies). . SD-WAN uses the Internet to build an entire business network or to add ‘thin location’ sites to an existing network. Startups in the SD-WAN space have successfully delivered a cost-saving business case to enterprises.
When cloud computing and multi-cloud emerged, many SD-WAN startups jumped at the task, recognizing that the cloud would require a new networking approach to operate optimally. At the same time, startups are recognizing security concerns and SD-WAN is starting to increase security. As a result, in the case of the cloud, Secure Access Service Edge (SASE) was created, which integrates SD-WAN and access security tools in the form of cloud hosting.
The hottest area in SD-WAN today is multi-cloud, which is an important indicator of how enterprises view network startups and also evidence that the opportunities for all enterprise network startups are disappearing in an instant. Of the 321 network startups, only 44 responded that they were multi-cloud users, while the remaining 302 companies were all VPN/carrier-Ethernet users and cloud computing users. Why did they jump into such a small target? This is because existing companies that acquired SD-WAN startups were dominating the field.
So where do companies today think network startups can succeed? It depends on how the company views risk and reward.
The area where companies today are most dissatisfied with their existing vendors is security.
295 out of 321 companies responded that they were willing to review strong security services from network startups, showing the highest level of acceptance. But “powerful” actually means something that is persuasive. This is because enterprises have a very low risk tolerance in the security area and the business risks of startups are considered very high.
A company’s security-related remarks made in 2006 are still valid today. “You want a supplier that you can sue if they don’t deliver on their promises.” While 295 companies said they would explore network startups in the security space, only 23 said they would be likely to work with a security-savvy startup.
Because the SD-WAN field can be easily expanded with security and acceptance of new network services, there are not many opportunities for startups to enter the field. Companies say they need more areas of their capital equipment budgets and agree that budgets for network equipment are primarily focused on data centers.
In this field, there have already been two new technologies that swept startups by storm, but both failed. These are software defined networks (SDN) and white box switching. The problem was that, from a potential startup’s perspective, neither had an attractive advantage: they lacked an innovative challenge to solve.
AI might have one of those.
Opportunities for Startups “AI Networking and IoT Networking”
Companies I’ve spoken with say that hosted, public generative AI won’t be able to transform their businesses. This requires AI that can analyze key business information that already precludes public cloud hosting due to governance requirements. They say that AI must be hosted in its own data centers, as that would also rule out public AI.
AI relies on clusters of servers equipped with GPUs, and hundreds of character prompts (queries in AI terms) can cause gigabits of data to move within the cluster to return results that are hundreds of characters in size. “This type of server-to-server traffic is called ‘horizontal’ traffic, and data center switches are already adding features to accommodate it. Will AI traffic models be different enough from traditional LAN switching to justify a new approach? Of the 321 people who commented on startups, 202 said they were considering or had already built AI self-hosting, and 144 of them said they would consider a startup to support networking.
What could accelerate startup interest in AI networking? One thing is clear: Companies are only now starting to consider hosting their own AI models. Network experts say that today the question of how to network a deployment is much more important than the question of which model to host and how to host and manage it. They also say that although a fairly universal way to approach building and connecting AI clusters is needed, they haven’t found one yet.
There is one interesting thing. Many companies (105 out of 144 considering emerging AI network vendors) believe that AI networking, container virtualization in the data center, and virtual networking on the WAN via SD-WAN should all be elements of new network and security models. The point is that it is being done. All 202 companies that responded that they were using or considering AI self-hosting said that they would like to review the service. But he also said that the winners among startups in this new combined space are likely to be independent startups “for a few weeks” before incumbent vendors catch them. But what good is a startup if it doesn’t innovate? There must be innovation to realize that reversal.
The company selected IoT networking as another interesting field.
The company believes that the combination of AI and the concept of digital twins, or computer models of real systems, will finally create a framework where IoT can shine. This will lead to the deployment of more common sensors, which will require secure and efficient mechanisms to connect these sensors. But above all, IoT networking requires a model that can connect to things like digital twins and AI insights.
The old age of networking connecting people to people and the medieval age of connecting people to applications and content are passing, 154 companies say. The ‘Age of Things’ is coming, which will create the networks of the future and grow some startups into giant corporations. And new-age security requirements may validate virtual networks, combining the opportunities of these two startups into something that companies say they can really jump into.
But that pesky qualifier “can” appears again. There are two factors that can ruin the startup parade.
First, it is what venture capital considers to be the optimal path to generating profits. Why jump into funding a challenging technology transition when you can jump on the social media bandwagon or the AI craze? Only 44 companies thought venture capital would fund useful networking startups.
Second, the competition is not with existing companies or other new companies, but with current upper-tier network companies or companies such as Broadcom. It’s easier to get from the minor leagues to the major leagues than to kick a ball in the sand. So while companies are ready, even eager, for the startup revolution, they are not waiting with bated breath. Perhaps in the next year or so, the greater threat of startups will lead to network innovation and reinvigorate the industry.
dl-itworld@foundryco.com
Source: www.itworld.co.kr
