Startup Nyansa Inc. today launched a SaaS-based IT network analytics service that can inspect, analyze and correlate wireline and wireless data to help large campus IT shops speed network problem resolution and create performance baselines that can be used for network tuning, gauging the impact of network changes, and justifying new network investments.
The CEO and co-founder of the company, which has raised $12 million in venture backing, is Abe Ankumah, onetime Senior Director of Products and Business Operations at Aruba Networks, who went on to become Director of Client Products and Alliances at Meraki. When Meraki was acquired by Cisco in 2012 Ankumah became Director of Cisco’s Cloud Networking Group, but left in late 2013 to cofound Nyansa with CTO Anand Srinivas and VP of Engineering Daniel Kan.
Nyansa (pronounced “ni-ans-sah”) prides itself on being able to gather and inspect wireline AND wireless data -- and even wireless metrics -- for use in performance analysis, and stresses that it doesn’t require agents on end user client devices and that it can correlate and summarize, in plain English, “data for every client network transaction from initial connection to application access.”
The meshing of the wired and wireless worlds is a good leg up for Nyansa, says Zeus Kerrvala, founder and principal analyst with ZK Research. “We’re living in a world where everything is connected, and in many organizations wireless is the primary connection, so being able to incorporate the two is critical,” he says. “Wired and wireless visibility is typically done separately and you end up having to correlate this stuff manually. There are probably some engineers out there that can do that, but most companies probably don’t have that level person. I think there is a real opening here.”
How it works
While the company’s Voyance service does not require installation of hardware appliances or client agents, customers will need to install one or more of Nyansa’s virtual machine-based “crawlers” that “extract and inspect all wired traffic from a mirrored span or tap port as traffic passes through network switches,” the company says. “This information is fused with wireless metrics gathered from WLAN controllers. Voyance currently supports Cisco, Aruba and Ruckus WLAN controlled environments.”
The crawlers look at all data but only forward performance metrics to the cloud for correlation by the Voyance cloud analytics engine, which is hosted in AWS. CTO Srinivas says that every gigabit of traffic results in about 500K bit/sec of information that is forwarded to the cloud for analysis. “Some of our largest customers are moving 3G to 4Gbps, so they’re seeing 2.5M-3Mbps going up to the cloud.”
The ability to see, collect and analyze both wireline and wireless data is key because Wi-Fi is “causing the biggest pain for the network ops guys because it has become the primary way people connect and it’s harder to troubleshoot those problems,” says VP of Marketing David Callisch, formerly of Ruckus.
“Within Wi-Fi we can see eight or ten metrics, everything from channel width, feature set, what access points are deployed and whether band steering is being used,” he says. “So if someone calls in complaining about a Wi-Fi problem, we can look at everything from the time they connect and every network service they go through to hit an application. It might not be a Wi-Fi problem after all. Maybe they’re not getting an IP address, or not authenticating to a server or the application is performing poorly. We can help guide IT ops.”
Early user Mike Fitzgerald, Network Architect and Senior Network Engineer at Brandeis University, says that kind of insight is valuable. Voyance “sees the conversations between the wireless clients and the underlying support -- the DNS servers, the DHCP servers -- so it will call out that, ‘This person seems to be having a bad experience, but it isn’t your wireless, its their DHCP didn’t work or your DNS lookups are slow or latency is off.’ So it is a great tool because you can go to one place and get a lot of root cause analysis.”
Those types of problems, Fitzgerald says, “would be hard to see unless you have a tool like this that can show you the correlation between all the moving parts. It has also helped us identify behaviors on our network that we didn’t know existed because they weren’t hard faults. Stuff just gets slow every once in a while and you’d like to know why. And now its like, ‘Oh look, when that stuff seems slow it is because this thing is happening over here,’ so it is great tool that way.”
But root cause analysis is only one thing Voyance was designed to address, Srinivas says.
Once installed, Voyance starts building a baseline of the environment, a profile that enables the company to address a range of other needs, including analyzing how network changes effect performance and end user experience.
“That’s one thing IT struggles with,” Callisch says. “They put in a new DNS server, and they don’t know if that change was good or bad. With Voyance, they can show what actually happened.”
Critical to that analysis is keeping the load the same so you’re comparing apples to apples, Srinivas says: “Let’s take DHCP as an example. You added another DHCP server and you expect it to reduce latency by X. You have to keep the load the same to see if it actually made a difference. Because we collect all this data, we can find periods of time when the load was the same for the before-and-after analysis.”
Because Nyansa collects all this data in the cloud, it also makes it possible to compare and contrast anonymized customer network analytics so companies can see how their environments stack up to those of similar sized operations in their industry, Srinivas says.
Sticking with a DHCP example, Voyance makes it possible for a customer considering DHCP configuration changes to see what happened when other customers tried a similar change. “We don’t make a recommendation as to whether you should do it or not,” Srinivas says, “but, without saying who they are, we can show you what happened when other people did it.”
Having the ability to compare and contrast also can help you figure out what projects to take on, he says.
“Let’s say network clients that cannot connect to DNS went from 0.5% to 0.3%,” Srinivas says. “The question any IT person would need to answer is, ‘Should I work on making that 0% or is 0.3% actually good enough?’” With Voyance you can see what the norm is for similar institutions and then focus your energy on efforts that matter.
That also has implications for network investments. If, for example, peer operations deliver better end user experiences, you can drill down and possibly use that information to justify, say, an upgrade to gigabit Wi-Fi.
Asked if any customers are worried about sharing this type of information, Callisch said none of the company’s roughly 25 customers have yet refused.
This experience sharing approach is “pretty cool,” says Kerrvala of ZK Research. “I read all the time about how IT isn’t that proactive. As a former IT guy I take offense at that. It isn’t that they aren’t proactive. They haven’t really had the tools to be proactive. You can’t manage a digital network with tools that were made for a different era, and that was what engineers really have had to do to date. It is getting more and more important to have management based on analytics because that lets you move to a more predictive model.”
Early user Fitzgerald at Brandeis says the university is just starting to work with Voyance’s compare and contrast functions: “We seem to compare pretty well. There are some places where we might be a little off so we’ll look at that.”
But the ability of the service to provide a bigger picture view is already paying dividends, he says. “We have a couple of buildings that have higher incidents of poorer performance and we didn’t think we had a problem in those buildings. No one is complaining, but now we have something to go investigate.”
In another case he says the service “showed us that one DNS server was getting hammered all the time compared to the rest of them and suggested we look into that. So we did and we found a firewall was pretty much using the same source port all the time so the load balancer was thinking it was all one conversation and was routing all its requests to one server. Its that kind of stuff where the value of this kind of product shines.”
Nyansa says it is targeting large campus environments with tens of thousands of devices in vertical markets such as education, manufacturing and healthcare.
Subscriptions to the SaaS service are available in 1-3 or 5 year contracts for different size buckets -- small, medium, large, x-large and custom.