According to many experts, the heavy networking and security requirements of implementing AI will require an overhaul of security architectures and tools – and AI-friendly tools may themselves need to. even be managed by AI.
“When it comes to networks, there are two types of AI: AI for networks and networks for AI,” explains an industry analyst. Zeus Kerravala wrote earlier this year. “The first uses AI to manage the network, and the second deploys a network to support AI.”
Network and security provider Aryaka touts its merits Unified SASE as a Service as the optimal architecture for companies that anticipate intensive use of Generative AIand as Kerravala predicted, AI itself will play a role in overseeing the implementation of the service.
“It’s the convergence of networking and security,” Klaus Schwegler, Aryaka’s senior director of product marketing, told us. “An approach designed from the ground up to have a unified policy, a unified way to administer, manage, orchestrate policies and have unified control over them when it comes to network security.”
Optimizing for AI…
To this end, Aryaka’s AI strategy has two sides. One aspect is adapting networking and security to AI needs, using three optional features for Unified SASE as a Service.
The first, AI>Perform, ensures that Network performance is optimized for AI workloads and applications. The second, AI>Secure, protects these AI processes by controlling access and stopping data leaks. Finally, AI>Observe gives users maximum visibility into their AI processes and network usage in general.
“This is crucial in an environment where real-time management and management network security are essential, due to the increasing prevalence of AI“wrote Renuka Nadkarni, Aryaka’s chief product officer, in a recent business blog post. “The ability to observe and analyze network performance in real time allows businesses to quickly identify and respond to potential security threats, ensuring more resilient and robust network operations.
AI requirements will reshape network architectures and procedures. For example, since each request to generative AI results in a unique dynamically generated response, there is no point in caching the data. However, because traffic to and from AI content servers will be massive in both directions, extremely low latency is a must.
The solution could be to geographically distribute the AI servers, a bit like SASE Points of Presencewrote researchers from the Orange group Usman Javaid and Bruno Zerbib a recent article on TM Forum.
“Future networks must extend cloud-centric architectures to the edge, bringing LLM closer to data sources, enabling low-latency inference, improving data transfer by processing data locally, while preserving privacy user data,” they wrote.
…and use AI to optimize
The flip side is using AI to extend network security and performance in ways that human-controlled processes could not. Schwegler explained how AI could, for example, detect unusual network activity that might escape human attention.
“Using AI tools to detect patterns and anomalies,” he said. “Traffic behavior that seems abnormal. … All of a sudden you have a spike in traffic, data transfer rates that you wouldn’t detect or too late to understand what exactly is happening, who is doing this.”
Spotting such anomalies is one of the several ways AI can strengthen cybersecurityaccording to the US Cybersecurity and Infrastructure Security Agency (CISA). Other aspects of cybersecurity potentially assisted by AI include the detection of personally identifiable information (PII) and participation in forensic examinations.
“There are patterns of behavior, trends that can be identified with AI much faster than any human deterministic machine learning model that can be written,” Nadkarni noted in a recent article. company webcast. “We also need to use AI to improve protection capabilities, secure access and secure all assets.”
The role of managed service providers
As a company that bridges networking and security, Aryaka sees itself as well-positioned to offer its customers and users services that will enable them to maximize their use of AI securely and efficiently. The target market is not only direct customers but also passing customers. managed service channels Also.
“MSPs with expertise in AI technologies will play a key role in spotting errors and ensuring the smooth integration of AI into IT workflows,” Nadkarni wrote in his blog post. “AI, in turn, will play a crucial role in improving real-time network security by providing advanced monitoring, analysis and error detection capabilities.”
Ken Rutsky, chief marketing officer of Aryaka, said in the webcast that adding AI assistance to network and security tools, as well as using these tools to focus efficiency and delivery towards the best possible performance of AI, will pave the way for a new phase of business opportunity.
“Our goal is to help our customers achieve it all: performance, agility, simplicity and security without compromise,” he said. “And as they move into these Gen AI applications, getting all that right will become more difficult but even more important and more rewarding.”
