blackboard - stock.adobe.com
4 ways to improve the network monitoring experience
The network transition from data centers to hybrid and multi-cloud environments makes the network delivery path difficult to find. Here are four ways to make that process easier.
Networks no longer reside only in on-premises data centers. Instead, many organizations host their networks through data centers and the cloud, complicating the network path. Therefore, it has become vital to understand the way data moves through the network.
Networking experts gathered at the AI Networking Summit Fall 2024 conference hosted by ONUG in New York City to discuss AI and its use in networks. Though there are many aspects to implementing AI successfully into networks, one important step is to ensure networks are resilient enough to handle the workloads, said Jeremy Rossbach, chief technical evangelist at Broadcom, during a session on enhancing network monitoring.
"None of your AI aspirations are going to come true unless you have a resilient network delivering those AI workloads," Rossbach said. "We still need to observe and monitor those networks to ensure your AI aspirations come true."
In Rossbach's session, he explained how network monitoring is necessary to ensure network resiliency within these complex and highly distributed networks. He discussed four ways to improve the network monitoring experience and untangle the complicated web of network devices and applications.
1. Embrace the new network delivery path
Rossbach explained that the network delivery path isn't as easy to follow now as it used to be. One reason for this is because networks are no longer confined to data centers.
"The network path does not look like it did when I was running data centers," Rossbach said. "Everything was in the four walls of your data center. You knew every packet, every user, every route."
Now, some enterprises don't even have a data center in use, moving to hybrid and multi-cloud options from cloud providers, Rossbach said. As a result, network teams have lost a lot of visibility into the network path, which has created operational blind spots. Teams must adapt and figure out their network's new delivery path.
"Embracing the new network delivery path is what we need to do. There's no getting around it," Rossbach said. "We need visibility to remove these operational blind spots into ISP and cloud providers to understand the entire end-to-end network delivery path for our users."
2. Work from actionable data
Data is unquestionably important to network operations. Rossbach compared it to cooking a dinner with missing ingredients -- recipes don't turn out correctly without all the ingredients. In the same way, network teams won't have proper visibility and insight if they don't have all the necessary data. This is especially true for AI tools.
"You're not going to get the right AI answer you're looking for if it doesn't have the data it needs to give you the answer," Rossbach said.
Though data collection can be tedious, collecting the right kinds of data only serves to help networks in the long run, Rossbach explained. It is also helpful for teams to reduce the scope of where they're collecting data from. Rossbach encouraged teams to start small in a vital area of the network -- in a certain packet or app -- and gradually move to bigger areas.
Ultimately, this data should help network teams understand what's going to happen in the network's future, Rossbach explained. Further, it enables teams to constantly pinpoint and adjust the network's reliability as AI workload demands change.
3. Trust but validate
Since many networks and network applications have moved out of the data center, they've become more complex with a data path that is more difficult to track. Therefore, validating the network delivery path as it reveals itself is important.
Validating the network delivery path from wherever the user is in the world and whatever ISP they're using helps network teams understand how users connect to an enterprise application, Rossbach explained.
"This is active monitoring -- being able to understand in real time what the network is doing, how it's behaving and continuously making improvements on the data you're collecting," Rossbach said.
Active and continuous monitoring leads to long-term network health. Enterprises need continuous validation to understand their network resiliency, Rossbach said.
4. Continuous improvements
Continuous improvements are vital to enhancing the network experience. Rossbach explained that teams move into this step when they understand the following factors and how they relate to their networks:
- Packet loss.
- Latency.
- Jitter.
- Capacity.
- Mean opinion scores for networks the company doesn't own.
- Unmanaged networks.
- ISP.
- Cloud providers.
As enterprises move into cloud networking, they must have insight into the cloud environment. This includes having cloud-first insight as well as combining cloud infrastructure network data with the rest of the collected network data, according to Rossbach.
"[It] helps us understand how the cloud network is performing as it relates to user complaints," Rossbach said.
Continuous improvements also help with understanding north-south network paths and east-west dependencies. By using a hop-by-hop analysis, teams can see what's happening at each stop along the way until they find the source destination.
"The topology of your network has so many dependencies that we need to be able to understand all those transactions across any multivendor technology," Rossbach said.
Nicole Viera is assistant site editor for TechTarget's Networking site. She joined TechTarget as an editor and writer in 2024.