Four months ago, I blogged about a new startup in the content delivery space called Instart Logic. Since then, I’ve gotten a lot of questions about how their technology works and how it compares to other CDNs in the market. These days, many find it hard to know exactly what kind of content delivery and acceleration services vendors offer, as many of them use different terms to describe their delivery services.
Instart Logic claims to have developed a new and unique approach to delivering web content faster to end-users. If the technology works as promised, this could signal a shift in the way web applications are delivered and it may change how the industry relies on edge caching, network acceleration, and full page downloads today. Instead, the industry could migrate towards newer types of web performance platforms that can make more intelligent decisions on how to order and prioritize pieces of an app or a site to send to a browser to improve user experience and diminish waiting time to first click.
Instart Logic’s approach also has the potential to better handle the higher levels of personalization we are seeing with dynamic sites and web apps, as well as the challenges that wireless networks are introducing into the delivery path. Instart Logic claims to be a full drop-in replacement for CDNs and the company clearly wants to compete with the likes of Akamai and EdgeCast. Whether or not they can do so at scale and performance is still to be seen. But keep in mind they don’t need to get to the size of an EdgeCast or Akamai to disrupt the market. Cotendo was only doing about $25M a year in revenue when Akamai acquired them, due to the pricing pressure Cotendo exerted on the market.
At a more basic level, Instart Logic is trying to solve several problems faced by content providers. The web is going wireless in a hurry with more and more users accessing sites from smart phones and tablets and they expect the same initial web site load times and rapid first-click response times as wired computers. They also expect personalization, social network integration, and other forms of dynamic data that are hard to cache.
The new generation of high-resolution displays on smart phones and tablets means that these devices can also display detailed HD photos. For visual segments like media and entertainment, travel and ecommerce, high quality images are absolutely critical to boosting engagement and ultimately revenue. Bigger images, more advanced personalization and more interactive content on web pages has resulted in rapid growth in the size and complexity of web sites.
The predominant focus to date of web performance companies has been to focus on wired era issues and reducing latency by caching pages and assets as close as possible to the end-user. The goal of this strategy is to minimize the number of lookups and round-trips back to origin servers. For mobile devices and wireless networks the focus has been to compress anything possible (images, JavaScript, and other static assets, etc.) to shrink the amount of bits that need to get pushed through a pipe with variable and unpredictable performance. Akamai has been promoting its adaptive image compression technology as a mobile solution in this space along with existing last mile asset compression technology.
Content owners say that if done incorrectly, some of these adaptations, including adaptive image compression, can have unpleasant side effects. For example, compressed images look particularly washed out and lifeless on Retina-class displays on smartphones. Worse still, some content owners are still loading mobile sites on larger screen tablets, subjecting users to really bad visuals.
Another tactic content providers use to speed up mobile user interactions is to reduce site complexity and quality themselves with simpler UIs, fewer images and less JavaScript. By minimizing the amount of data required to load a page on a mobile version site, publishers can load pages faster and allow users to interact more quickly with sites on mobile devices. But the downside is that users are less likely to click or spend time on boring, less dynamic, visually uninteresting sites that look like our desktop web sites from the late 90’s, albeit on a smaller screen.
Instart Logic claims to solve these problems by allowing content providers to deliver visually appealing and complex web experiences for mobile but with fast response times across all screens. Their system does this by intelligently prioritizing the order of web page elements (or pieces of those elements) delivered to a user’s screen. It does this by analyzing the different parts (images, HTML, etc) of a web page and studying how and in what order browsers load these parts. Instart Logic’s technology then divides the different parts of the website into much smaller fragments. Then it makes intelligent decisions about which fragments to send first and which to send later, all the while streaming first high and then low priority pieces into a browser continuously.
Instart Logic claims this allows users to quickly see and interact with web sites even with big images and heavily dynamic content. It almost sounds like the early days of streaming video, so here’s a bit more detail on the components of their system and how they work together to accomplish the speed gains.
- The NanoVisor – This is a very small JavaScript virtualization layer that the company injects into the website as it goes through their service. This virtualization layer sits between the browser and web application and acts as a client for the client-server architecture that Instart Logic constructs from the user’s device up to their cloud based web application streaming network. The client-server architecture is the same as what RealNetworks and Adobe built for their own video streaming and windows application streaming technologies. But by using JavaScript and modern HTML5 technologies, Instart Logic does not require a user to install any software or change browsers.
- The NanoVisor has two key functions. First it captures key information about how web applications load and which pieces are the most important to load first. It also acts as the receiver for a web application stream from the cloud part of their service. NanoVisor works on all popular HTML5 compliant browsers and across all devices. While Instart Logic primarily built its solution targeting the fast growing mobile and wireless device segment of users, the NanoVisor also works well on traditional sites used with laptops and desktops. While it could replace the existing use of a CDN it can co-exist with FEO systems or with in-house performance tuning because the NanoVisor and Instart Logic’s web application streaming network deliver performance on top of existing web performance changes.
- AppSequencer – This is a cloud-based intelligence engine that does the analysis of the various website load patterns based on data collected through the NanoVisor. The AppSequencer will look at how a web application is loading on devices and build a profile for that application. The AppSequencer analyzes which images, HTML and Flash components need to render quickly for the user to see and begin to interact with a Web app. Then, the AppSequencer will fragment the images, Flash files, and HTML into smaller pieces. It then streams those components to the NanoVisor in order of priority, pushing the most important bits first. The AppSequencer can currently stream images, Flash, and dynamically generated HTML but they tell me they are working on other content types as well.
- The Network – Instart Logic has launched with 30 PoPs covering many of the major Internet peering points and their architecture is a hybrid with a mix of physical and cloud based locations. That’s less physical coverage than you get from Akamai or another very large player, but Instart Logic’s whole philosophy is to move away from the old-style big iron CDN approach because they believe the Internet backbone is more stable and much faster than when companies like Akamai came into existence. The company states that their network takes web publishers close enough to the vast majority of devices and users to deliver what is effectively the same origin-to-edge-of-network-performance for static assets. And they point out the problem is now on the wireless last mile beyond the reach of the existing CDNs systems.
So how would a web application load process look different running through Instart Logic’s web application streaming network compared to a CDN? A typical web app would pull all static components from a CDN edge cache only after loading all required HTML upfront, wasting precious time. It would then start processing all of that and then load up the site. It requires more waiting and much more data download just to get an initial display and allow interactivity like scrolling and clicking by the user.
In contrast, a web app served through Instart Logic’s streaming network would load the non-unique HTML up front, then the NanoVisor, followed by CSS and JavaScript components. Next, Instart Logic would load a portion of the first image a user should see on a page. It would then load the rest of the HTML and load up a portion of the other key images required for a user to get a full flavor of the page. At that point, the application is ready for user interaction. Instart Logic will subsequently stream down the remaining bytes of the images or Flash files in order of importance in the background.
To be clear, Instart Logic’s provides much of the functionality as a CDN. It has the capability to cache static assets close to the end-user and can deliver first and middle-mile network acceleration from the origin server to the edge. Instart Logic optimizes TCP/IP at the network layer to speed up HTML transfer. They also tightly manage connection settings between PoPs to save time by keeping connections open longer. But the real secret sauce is what happens in the last mile, which is where the client-server connection between the NanoVisor and the AppSequencer lives and where today’s CDNs have the least to offer. The full package is a replacement for CDNs (and ADNs), the company claims.
When I ran the concept by those who build out today’s leading CDNs, they said it definitely should work if executed well. The segments Instart Logic appears to be targeting are media, travel sites, game sites, and e-commerce sites. As I mentioned in my last post, I could see a noticeable difference in web performance in site load and response times, when Instart Logic’s execs demoed the product for me. Even more interesting, Instart Logic showed me before-and-after side-by-side demos on sites that aren’t even their customers, using live production sites pulled in through the Instart Logic service.
It’s too early to know whether Instart Logic’s technology can represent a true paradigm shift in the industry. But of all the startups in the space right now who are trying to solve some of the hurdles of streaming web applications to mobile devices, Instart Logic seems to be the most interesting. The company has recently announced its first set of customers with some case studies on their website.
Over the past year, Yahoo! has been making a big push into the video market, (licensing Saturday Night Live videos from NBC Entertainment) as well as creating and syndicating original content. So I’m very pleased to announced that Erin McPherson, VP and Head of Video at Yahoo!, will be the keynote speaker on day two (Nov. 20th) of the Streaming Media West conference. Come hear what Yahoo! sees as the future of online video consumption and learn more about their video licensing and content creation strategy.