Vasona: Reimagining Network Monitoring, Enhancing QoE for Operators

As telecom operators strive towards delivering the best data-driven quality of experience to their end customers, precise network monitoring can deliver network performance improvements for the operators and this is where Vasona Networks is delivering its innovative products and solutions to help operators do more with their existing network investments.

John Reister, VP, Product and Marketing at Vasona Networks speaks with Zia Askari from TelecomDrive.com about the core focus areas for the company, its focus on India and how it is creating innovative solutions around the Mobile Edge Computing space.

Tell us something about your key priorities today in the market? 

We look at the congestion in the mobile radio access networks and we observe the traffic and when we look at it we see the demand fluctuating and the capacity fluctuating and the intersection between those causes poor experiences for the consumers.

When there are constraints on capital investment, spectrum, these things become even more critical. But to be honest even where you do have capital spending, the traffic growth and the demand growth rates are so insanely high that everybody is scratching their head and trying to figure out how they are possibly going to meet this demand within the next few years.

So our main focus is, at Vasona, we go inline in the networks and we can perform traffic management and by doing that we improve the quality of the experience for the end users, make browsing work, the page load faster, we make video start time go down, the stall times are reduced and hence, social media works more quickly.

Also during busy hours, you can fit more of those services through the network so you actually get more out of your investment. It’s true for the investment you have already made and for investments you will make in the future.

Tell us about the focus that you have globally, the big geographies that you are targeting?

The India sub-continent is a big focus for us. We have a local team here and we have an unannounced deployment, we have been doing an extended pilot over a number of months evaluating the network and showing the benefits of our technology and then we are also deployed in Latin America and we have an announced deployment with Telefonica in Europe.

Today, if you see there is a major transformation happening - telecom operators do not want to be known as service providers, they want to be known as experience providers of their customers. In that scenario, yet they are looking forward to a better monetization strategy, how can they monetize in a better manner? In that scenario how can your solutions help them?

There are a few things, this area where we sit inline between the RAN and the core, it is a part of an industry initiative called Mobile Edge Computing, where it brings Network Function Virtualization concepts towards the edge of the network and so there are a few aspects to it.  Our product is available immediately with the ability to do a MEC use case called "video guidance".

Video guidance is where the mobile operator can inform the content provider as to the current capability of the cell that a user is in as to what bit rate they can stream video at.  In other words, if the user is in a congested cell, the operator embeds into the get-request messages that go over to that CDN, a message that says they can only sustain 500 kilobits for that user. And then another user who is in a less congested cell, say 800 kilobits.

The idea is to allow the video to get to the best rate, the best sustainable rate as quickly as possible that means the user gets a better experience.  It also means the network is neither under nor over utilized, it’s just right. The benefit is that now the operator is working collaboratively with the CDN, they are working together.

The operator can bring a small tiny piece of information about the state of the network where that user is, and then it’s being delivered to the benefit of that end-user who is actually both of their customers. It allows a more collaborative relationship between them.

On top of that MEC, part of the purpose is to be able to host applications that can run on this platform, it’s a flexible and extensible platform, and allows them to do everything from IoT and caching and communications with lots of devices, very low latency types of services and it leads you to 4G and then to 5G, the effort is part of the interest in 5G, it’s about giving the operator more tools and more levers to pull to create customized and targeted capabilities and services.

In India we are having this major issue around call drops, all the operators are facing this issue in various parts of the country. So, in your opinion what is the best way out for an operator? There are multiple solutions. Some people say that cloud RAN can solve it, rerouting of the traffic around those different sites would solve it. What is your opinion on this?

Where it is 3G, those are circuit switched calls and Vasona is mainly focused on the data and packet functions, however we do influence the control plane a little bit and part of what happens in voice is inability to establish a bearer. The challenge there tends to be how many bearers are enabled and how close the cells are together which causes interference and excessive handovers.

But as you go to 4G and you have VoLTE calls, and VoLTE has a similar problem of call drops, then Vasona can directly influence those voice packets.  If there are too many routing hops between the origination and termination point the latency goes too high and you end up with a very poor call – you can’t have the conversation.

One of the use cases for Vasona's MEC solution is to improve and manage the VoLTE call quality and there are certain elements of the VoLTE delivery system that can be housed in the mobile edge that can reduce the latency and improve the quality of those calls.

For 3G IuCS, the circuit switched voice, Vasona is on the IuPS, the packet part, so we don’t do anything to directly influence today’s dropped calls made on 3G or 2G, but on tomorrows VoLTE, absolutely as it’s packet based.

How many customers do you have globally and what kind of growth are you looking at?

The growth is enormous. We are a small company based in California and our customers don’t typically allow us to announce. We don’t reveal our number of customers, but we are contracted for over 100,000 cell sectors and hiring operations people as fast as we can.

We have about 2/3 of those installed and obviously we are selling more. India is a primary market, it’s a large market and the mobile QoE challenges are significant and the spectrum is scarce and the customers have a great appetite for video, there is a real interest in video in India.

In India what kind of plans do you have in terms of going to the market, do you have any plans to partner with certain key organizations?

We have a direct sales team, it is a "high touch sale” meaning we provide a lot of consulting during the sale.  In terms of partners, the quality of experience is relatively hard to measure, so we team up with partners for drive testing and things like that. Another area is analytics.

Today, it is very hard to tie the user experiences to the specific cells causing those experiences. A lot of times measurements are looking at thousands of customers’ traffic and can determine if the operator is delivering a good or bad experience but they can’t go back and tie it to the cell.

Or there are measures that will look at a cell and say it seems to be overloaded, but they have no idea how many customers are affected, are they experiencing video stalls, etc. Vasona allows them to put that together. We get all that metadata, but we are not a big data analytics company, so we are interested in working with partners where we can export that data.

We have a product interface called SmartCONNECT, and we export that data to third parties. Most operators just bring that data in-house and they store it in big Hadoop databases, but we think they could benefit from third party analytics companies so that’s something we are looking to do. Finally, we do sell direct but as we start to scale and need to install more and more systems, I suspect we will be looking at local fulfillment partners.

By when do you think you will be in a position to look for those types of partners in India? 

In the next 3-6 months, we are deployed in a commercial pilot situation right now.

When do you think that will go live?

It’s been live for months; we are transitioning to commercial scale.

How many end customers would be touched by your solution?

About 1.5 million currently in India, which is a big number, perhaps small for India, but still a big number!

Today, what kind of challenges do you foresee in terms of the space that you are in as a vendor?

We sell to the operators, and historically, the operators and vendors are divided between the RAN and the Core. RAN people are all about power and SINR (signal to noise and interference ratio), spectral efficiency, etc and you start talking to them about packets and it’s the wrong language.

Core people don’t focus as much on the cell and are all about the packets, and switching and policies, PCRF — they are less interested in cell congestion. What’s happening with this industry MEC initiative is that the intelligent edge is moving towards the RAN and is now falling somewhere in the middle organizationally between the RAN team and the Core team. That’s a challenge for the operators because the organizations are divided and it’s really only the CTO who has that visibility across, so it’s a challenge for vendors like Vasona to sell into that environment.

We sell to RAN teams, because we make the RAN more effective but the measures of success are things that the Core team is currently more focused on. It requires a multi-functional approach at the operator to understand the problem and understand the solution to the problem.

Is there something about the big advantages that your solution brings for the network operators?

Up until recently there were a couple of available solutions to tackle Quality of Experience and video in particular. One of them was transcoding or transrating and it involved taking the video and decoding it and re-encoding it at a lower bit rate.

The other solution was to cache transparently so when a user uses a piece of content you store it and then if another user an hour later requests the same content, you can then send it down immediately without having to go over to the transport and through the core.

Both of those solutions relied on the content to be unencrypted, but now 90% of video is encrypted so these solutions no longer work. Vasona is able to provide our QoE benefits on encrypted traffic and that’s an advantage of Vasona. Other approaches are to try to do this function from the Internet side, where the Internet plugs into the core, at the Gi or SGi interface.

But from that location, you cannot figure out which users are contending for the same cell resources, from that location you might see a million users but you don’t know out of that million users that ten of them are in the same cell sector fighting for the same resources.

Also, with the traffic fluctuating so rapidly you have to respond very quickly, you can go from a cell that is working fine to a cell that is jammed up in 300 milliseconds. I call it the speed of congestion, sort of an oxymoron. Congestion is like a car accident, the road can be moving along well and within moments it can get backed up.

In less than 300 milliseconds, a cell can be completely full--packets are being dropped, the queues are full, so you have to have a very rapid response time.  Because Vasona has put the detection of the problem and the resolution to the problem together in the same platform, we can generally respond within 10 to 20 milliseconds.

Other solutions that respond in 2, 3, 4, 5 seconds — that demand spike is over and you may very well be in a valley where the traffic has backed off or you are in between video chunks, and now you may be trying to do something when you are at a lull and it can cause more damage.

What does your competitive landscape look like?

There are some DPI solution companies, but again those solutions sit in the wrong place in the network on the Gi or SGi interface. Sometimes they try to connect a probe to those DPI solutions, often going through the PCRF, and the problem there is, I have heard, response times of several minutes, lengthening the time detecting the congestion and the time of taking action, you must respond well under 100 milliseconds to be able to tackle congestion.

Those are the main competing alternatives, the caching companies, transparent caching just doesn’t work anymore. You can try non-transparent caching but that’s a whole different business model that’s where you are going to participate in the certificate exchange, the TLS, security exchange between the user and the content provider and moving that out into the edge of the network is a fairly substantial, far off, future.