Virtualization Technology News and Information
Q&A: Interview with Instart Logic Talking about the Web Application Streaming Network

Peter Blum heads product management and product marketing at Instart Logic, which unveiled its new Web Application Streaming Network June 13, the week before the O'Reilly Velocity conference where the world gathers to talk about the latest advances in web performance and optimization solutions.  I've previously spoken with Peter when he was at Citrix, where he was senior director of product management and marketing for the XenClient product.  So I thought it appropriate to reach back out to him to see if I can find out more about what his latest venture is all about.

VMblog:  How does Instart Logic differ from existing web delivery systems?

Peter Blum:  Instart Logic has developed a radical new solution, built from the ground up to address the challenges of the new realities of the mobile and wireless web, called the Web Application Streaming Network. It works by dramatically reducing the amount of up front data transfer required to display and run a web application.

It uses a novel new approach based on analyzing how browsers load and interact with web applications. It divides a web site or application into fragments and re-orders those fragments into a stream, sequencing the most important ones first. The system then adaptively streams the application to a user's web browser. As a result, with only a partial download of the application, the service allows users to view and interact with the application dramatically faster than with traditional approaches. The remainder of the data then downloads in the background delivering the exact experience a web publisher designed.

VMblog:  So is Instart Logic a piece of software, a hardware device, or a service?

Blum:  Instart Logic provides a cloud-based service. So with no hardware or software to deploy or code changes required customers can easily start using the solution with some simple updates to their DNS records. They just redirect their end users through our system and we take care of the rest automatically.

VMblog:  How is Instart Logic different from a CDN (Content Delivery Network) solution?

Blum:  Content delivery networks are cloud-only solutions that work by keeping exact copies of frequently accessed images, scripts, and other web site components "cached" closer to users in ISP edge networks. They were originally developed 10 years ago when latency in the core of the Internet was the main bottleneck to fast website delivery.

CDNs were not designed to address the new challenges of highly personalized web sites and the network bottlenecks that now exist on the wireless last mile networks between the ISP edge networks and end users. As a result CDN approaches stop at the edge of the Internet and don't address modern web delivery challenges. Further, CDNs don't have an understanding of what parts of a web site or application are most important and as a result require sending all data up front before a browser can display it.

VMblog:  So how does Instart work exactly?

Blum:  Instart Logic's Web Application Streaming Network works by dividing web sites and applications into smaller fragments and then intelligently streaming the most important parts of the application to the browser first. There are two unique components to the Web Application Streaming Network: one component in the cloud and the other on the client side.

VMblog:  Tell me about the client.  I understand you inject a hypervisor into the browser?

Blum:  Yeah, I didn't quite leave the virtualization world behind when I left Citrix and XenSource. Instart Logic gathers client side intelligence with something we call the HTML5 NanoVisor. This is a thin virtualization layer, with a footprint of only about 30kb that sits between the device's web browsers' various APIs and the running web application. The NanoVisor is injected into a user's browser at the beginning of a session and is completely transparent to the user. It is written in JavaScript so its works with any HTML5 browser.

It does two main things. First, it allows us to watch how a browser executes a web site or application and it monitors which assets for that site are requested and in what sequence. Second, the NanoVisor reports back this information to Instart Logic's cloud component, called the Personalized AppSequencer.

VMblog:  And what is the AppSequencer?

Blum:  The AppSequencer is the cloud part of the service that analyzes data returned from the NanoVisor and performs the actual fragmentation of web site assets, including image files (JPEG, PNG, WebP), HTML and Flash. I'll give you an example for image files, The AppSequencer can break an image down into smaller fragments. One set that gets delivered up front to help get the initial render of the page done quickly with minimal data transfer and then the remainder of the fragments come down soon after to deliver the full visual experience.

VMblog:  I see that Kongregate is one of your reference customers. They have a bunch of flash video games. Can you tell me how you work with flash files? This could breathe new life into a lot of flash games.

Blum:  For Flash files, the AppSequencer understands the inner workings of the Flash or SWF format. For example, it understands the various embedded images, sprites, audio, video, and code components within a Flash file. So it can take what are usually these really big files and divide them into smaller fragments. The system will then automatically figure out the minimal set of fragments that are required to start playing the game and send that down up front. The rest then comes down in the background as needed. Since Flash games tend to be so big you can get a really amazing reduction in download and start to play times.

VMblog:  What is your approach to dynamic HTML pages?

Blum:  Well it turns out if you know what to look for you can find a lot of non-unique HTML across dynamic HTML delivered to different users. With traditional web delivery systems they resort to only using network acceleration on dynamic HTML since it can't be cached and must be generated per user by the backend web servers. So a web browser makes a request and then sits doing nothing for sometimes seconds while this all goes on.

Our Personalized AppSequencer will automatically examine these dynamic HTML pages and identify code and assets that don't actually vary across users. Then in the future when a browser asks for a dynamic HTML page the Instart Logic system can immediately start sending this non-unique information from a server near the end user while the dynamic part is still being generated on the backend. So now the browser is not sitting doing nothing anymore we keep it busy. And that lets the page load up much faster.

VMblog:  So tell us if you will, what's next?

Blum:  We are just getting started. While traditional web delivery systems are backend or cloud only we have this amazing new architecture with client side virtualization running in a browser. Then you add in the application intelligence in the cloud and that is going to open up some really interesting new possibilities not only around increasing load performance but also in other areas that can improve user experience. All of a sudden we have a footprint on both sides of the network connection so that opens up some big areas for additional optimizations.


Thanks again to Peter Blum for taking time out to speak with VMblog.

Published Wednesday, June 19, 2013 7:13 AM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<June 2013>