|© 2016 Black Swan Telecom Journal||•||protecting and growing a robust communications business||• a service of|
|Email a colleague|
Predicting the uptake of new technologies is about as accurate a game as throwing darts at a wall blindfolded.
Vendor press releases, blogs, and PowerPoints contain a good measure of bias. Then there are the “independent” analyst reports with their hockey stick adoption curves.
But tracking the activity of high-tech OEMs is a legitimate technique for spotting trends. For instance, camping outside Foxconn factories in China should be a fruitful place to do electronics industry research because Foxconn, the world’s largest electronic contract manufacturer, makes the iPhone, iPad, Kindle, PlayStation, and countless other devices.
Well, the equivalent of Foxconn in the network management space in WebNMS. For a decade or more, WebNMS, the telecom OEM software side of Zoho Corporation, has led the market for commercial frameworks used to build either Network or Element Management Systems (NMS/EMS) In fact, there are more than 800 WebNMS deployments each quarter and they range from applications that manage $25 cable modems to multi-million dollar optical long haul switches that number in the dozens.
I caught up with WebNMS through Eric Wegner, the firm’s Marketing VP and regular contributor to the B/OSS blog site. I was curious to get Eric’s view on a couple key trends: M2M and analytics, especially as those technologies intersect with network management.
|Dan Baker: I know one of the areas you’re increasingly serving is Machine-to-Machine (MTM) or the Internet of Things market.|
Eric Wegner: Yes, Dan. As part of a wider communications package or family plan, M2M is very attractive. For instance, an M2M service like medical or security monitoring in the home has a high perceived value: it’s a sticky service because consumers fear losing coverage if they switch their mobile broadband provider.
The M2M applications themselves don‘t carry a lot of bandwidth services like video. In fact, the messages are rather small — like SMS. Where the complexity comes in is where you have a huge number of devices out there: medical diagnostic gear, industry process controllers, and the like.
And today many of these devices don‘t have a good means to communicate back with the manager of these devices. That’s where our WebNMS comes into play: we offer them a path to do fault monitoring, performance management, and security.
One of the problems with M2M as a software market is that it’s highly fragmented. It’s everything from cell tower management and medical devices... to solar / wind power remote management and grocery stores monitoring the hydrogen in their refrigerators.
And in many cases the people involved are end users, not developers or system architects so they don‘t have the internal capability to use our framework. For example, one customer needs remote management of IT infrastructure on the oil rigs sitting out in the Gulf of Mexico. Now the people who run those rigs are more system admin types, so we need to customize solutions for them. And because we can’t afford to address the wide array of M2M uses, we often get systems integrators involved.
But as I say, the biggest issue in M2M is the explosion of devices because the amount of performance metrics grows very quickly.
|Now the M2M market is a far cry from the carrier-grade business you do with carriers, OEMs, and network equipment vendors like Alcatel Lucent.|
Yes, we are serving a wide array of customers in long haul optical, metro-optical, mobile backhaul, satellite, and wireless access. And that market plays to our framework’s strength in helping clients build custom network/element management solutions.
Now in that space, we’ve recently taken steps to speed the design-to-deployment cycle. We now build APIs to the framework to enable things like SDN (Software Defined Network). Likewise, we have a Carrier Ethernet “feature pack” that smoothes the path to using the MEF standards. These design aids mean customers don‘t need to build everything from scratch.
|I saw press release announcing WebNMS’s support for the Apache Hadoop architecture. What’s that about?|
Yes, we figure Hadoop is the next big thing so we’re gearing up for it.
A satellite application customer is looking strongly at the WebNMS and Hadoop combination for their project. It’s a bandwidth optimization use case involving satellite-to-ground comms: gobs of data and plenty of latency/jitter issues to sort out.
One of the beauties of Hadoop is you can distribute the load across, say, a cluster of 20 commodity CPUs instead of having one huge database and hardware.
We tested Hadoop’s performance in one of our largest deployments: 7 million cable modems stored in a relational database. When we compared the Hadoop application to a traditional one, Hadoop boosted query performance by a factor of 10.
We feel high-end fault management is perfect for Hadoop. That application involves correlating faults from neighboring devices or comparing the current fault with similar fault types in a historical store. So if you have several terabytes of data coming in and you need to analyze that in real-time, the Hadoop architecture can handle that. In fact, the Hadoop ecosystem supports lots of new real-time databases.
|What are the drawbacks of Hadoop?|
The technology is strong, but will people adopt Hadoop in sufficient numbers? That’s the key question. For companies who have a strong relationship with the large hardware firms, it will certainly be a fairly long adoption curve.
But for companies who are looking for value, the cost savings are compelling: in the order of $50,000 versus the low millions of dollars.
|On final question, Eric: what’s your opinion of search engines that explore unstructured or machine data, such as Ontology, Splunk and HP’s Autonomy? Will they gain traction in network management applications?|
Actually Splunk’s method of searching syslogs is similar to what we do with our Event Log Analyzer.
For troubleshooting, devices spit out syslogs -- raw messages telling you what the device is doing. So if something bad happens, the NMS sends an alert. And if you need to look at all the gory details, that’s typically where the syslogs come in.
Ultimately this is a design choice. WebNMS could pull in the syslogs and convert them into alerts. From a telco point of view, then, a syslog alert is an alternative to alerting with SNMP or another protocol.
So far, we haven‘t seen Splunk in network management. My guess is that Splunk’s strong suite is in enterprise compliance reporting and security. WebNMS also has a few large data center enterprise customers. One of them is a large bank who uses our Event Analyzer to check syslogs for security.
|Thanks for getting us up to data on, Eric. It’s highly interesting perspective.|
Copyright 2013 Black Swan Telecom Journal