eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
This week, high performance network vendor, Arista Networks, held a digital “Cloud Innovators” event. The company has used events like this before to make some major product announcements and this event was no different, as the company announced its network data lake, or NetDL in product terms.
NetDL is the next step in the evolution of the Arista platform and addresses the rise in artificial intelligence-driven networking.
Arista was a pioneer in software driven networking
Arista came to market as one of the first vendors to move to a centralized software model where its operating system, EOS, was decoupled from the underlying hardware. Arista products were built with a centralized in-memory state database making networks much more reliable and programmable.
The second phase of Arista evolution came in 2016 when it released its network database, NetDB, which was a consolidation of all data from Arista products. This includes alerts, flows, packet information, control plane traffic and device state streaming.
NetDB created a single platform for application and ecosystem development. Since its inception, Arista’s differentiator has been a single image across all its products, which made the creation of NetDB possible.
Almost all other vendors have a separate operating system for the WAN, WiFi, campus and other places in the network. Creating a network wide database would require ingesting the various sources of information and normalizing it, which would require a tremendous amount of heavy lifting
NetDL powers the next wave of Arista innovation
NetDL takes NetDB and adds a number of other data sources, including Internet and systems information and data from third parties such as Zoom, Zscaler, VMware, Microsoft and Palo Alto Networks.
This gives Arista a much broader set of data to process, analyze and derive operational insights from. NetDL can be thought of as a “single source of truth” for network data that enable better forensics and analytics for threat hunting, network packet brokers, network detection and response, performance monitoring and application experience monitoring.
All this data has existed for years but there was a missing link that made interpreting the data possible today, where it wasn’t in the past and that’s the rise of artificial intelligence. NetDL is a massive amount of data an AI requires data to be effective. In data sciences, there’s an axiom that states “good data leads to good insights” and that’s certainly true but it’s also a fact that partial data leads to incomplete insights.
The use of AI with NetDB might trigger some interesting information about the network but would be blind to the impact of the findings on cybersecurity, application performance and others.
AI plus NetDL enables faster threat hunting and app performance monitoring
With siloed networking and security, threat actors are able to breach a network and move laterally slowly, which keeps them hidden in plain sight as there is too much data to analyze manually. Even if the breach is detected, finding the source can be difficult, if not impossible as tracing the breach to another part of the network requires correlation across a number of systems and by the time its found, data has likely been exfiltrated.
NetDL sees everything across the network and security systems and enables the correlation of even the smallest anomalies, which can be used to find the source of the breach and automate the reprogramming of the network to minimize the damage.
Similarly, a Zoom call that falls under the threshold of acceptable quality could trigger an inspection of the network to troubleshoot the problem and isolate the problem. In a recent AIOps study conducted by ZK Research, we found that some engineers spend half of their time troubleshooting the network.
That’s because network troubleshooting can be very difficult, particularly in this era of software defined everything, cloud and mobility as a problem in one device can have a ripple effect and cause problems elsewhere. AI applied to NetDL can quickly see there’s an application problem and then quickly identify where in the path of traffic the problem emanated from.
Autonomous networks is on the near term horizon
Another benefit of NetDL is that it enriches Arista’s autonomous virtual assistant, AVA. The combination of AVA + NetDL can shift network operations from a reactive model to one that’s proactive as the virtual agent can watch the entire network 24x7x365 and use AI / ML models to predict when a problem will occur. AVA can then guide network operators through fixing the problem before its business impacting.
The partners announced this week should be thought of as a starting point for NetDL. During the event, Arista hinted at others coming in 2022 and will including a range of companies including other strategic partners and ISVs to deliver market and customer specific or application specific insights.
The IT industry has existed in silos for decades, but that model no longer works as applications are now cloud delivered and users are on the move. NetDL brings together a “full stack” of data to let the network understand not only what happened but why it happened.