Technology:
PacketZoom offers a new hybrid of mobile networking service - designed specifically for native mobile apps. Using the service is easy since we transparently proxy HTTP/S requests through our network stack without requiring any client or server side code changes.
At the heart of the solution is the PacketZoom protocol stack. Written on top of User Datagram Protocol (UDP), it operates completely in user space, with no special kernel facilities required. This means that it can work on all mobile devices in the world today.
PacketZoom's protocol downloads static and dynamic content faster. We reduce the number of round trips, the HTTP/S protocol overhead, and eliminate slow starts as well as aggressive backoffs that are characteristic with TCP. The result? Maximized use of available bandwidth, which means faster content downloads from the cloud.
PacketZoom's protocol significantly reduces the number of connection failures. Our protocol can seamlessly handoff a session across a network break even when the client IP address has changed (e.g. moving from wifi to 4g or crossing a tunnel in a moving vehicle). The result? Minimum number of disconnects and transfer failures, which means a more reliable and engaging mobile experience
Faster Startup
Latency, the time it takes for a packet to get from a server to device, is the number one speed killer. PacketZoom reduces the number of roundtrips to the server so that roundtrip latency has less chance to slow things down. The initial payload is determined by type of connection and variables that are specific for the given request, so if you're on a fast LTE connection you get more data upfront than if you're on flaky 3G or 2G connection.
// TL;DR
Fewer round trips per file and we use bandwidth hella efficiently.
Intelligent Packet Loss Handling
Mobile networks routinely drop a small percentage of packets because of the very nature of cellular and WiFi technology.
PacketZoom’s protocol has a built-in mechanism for actively monitoring the network condition as experienced by the receiving device. This allows us to react more intelligently to dropped packets. For example, a handful of dropped packets, which could decimate bandwidth usage over TCP, are far less destructive because of the active feedback mechanism from the mobile device to the server.
In contrast, TCP, has baked in 40-year-old assumptions about what a dropped packet means: TCP always assumes that a dropped packet means that a router along the way is experiencing congestion. There's no built-in feedback mechanism to allow it to judge the condition of the network apart from simply waiting for an acknowledgement of a packet from the receiving side. TCP acts aggressively to prevent router congestion in the absence of high quality information about what the receiving side is experiencing.
// TL;DR
Relying on 40-year-old assumptions is kinda dumb, so we actively monitor network conditions.
Seamless Connection Handling
PacketZoom creates a unique clientID as the destination for data packets, compared to legacy protocols that use only the device's IP address.
Treating the device itself as the destination allows PacketZoom to easily handle the intermittent nature of mobile connections in an intelligent, fault-tolerant way. So as your device moves across networks, any ongoing transfers can continue seamlessly something that is simply impossible with old-fashioned legacy protocols.
// TL;DR
We expect solar flares or tunnels to mess things up and built our protocol for the real world.
Architecture
Our technology is available in the form of an SDK. We’ve worked hard to make it very easy to integrate, giving you full control over the integration through our dashboard and made it fault-tolerant and very safe.
Our server was implemented as a set of virtual PoPs (point of presence) that run our proprietary software on any standard public cloud infrastructure (such as Amazon WS).
Simple Integration
We've designed PacketZoom to make it as easy as possible to use. Technically, you could integrate our SDK without a keyboard. You add our framework, copy our init, and that’s it. No need to modify every content request, or install anything on your servers.
No server or infrastructure side changes are required either. You can keep using your existing Web CDN service and should benefit from the PacketZoom addition in the last mobile mile.
Lightweight
Our SDK memory footprint has been carefully tuned to be as small as possible. Proprietary buffer handling techniques on the device take advantage of the the nature of hardware on new mobile devices. Result, there is no receive window. There. Is. No. Receive. Window.
Fail Forward Technology
Before the first request that uses our protocol, we check to make sure that everything on our end is healthy and fast. That is, we start in a virtual “failed” state and only start handling traffic once an all-clear is received. Result, no waiting for timeouts. If anything goes wrong with our systems, your users won’t be impacted so the worst case scenario is that your app is the same speed as it is today.
Dashboard Control
You always have full control over your app. Among other functions our dashboard lets you (a) turn the service on/off (b) define content filter to determine what type of content to handle, (c) what percentage % of traffic to handle and (d) control client and server cache sizes and purge control.
Caching
While the main benefits of PacketZoom’s technology are around the protocol it does cache static content when possible. PacketZoom offers (a) application cache as well as (b) server side cache. Both of them are controlled in real time through the Dashboard. Note that PacketZoom service uniquely allows you to trigger instant purges on the client side. Not just on the server side.
Each time PacketZoom serves content from either application or server cache, the content is served faster and your CDN usage/bill decreases
Benefits
Designed from the ground up for a mobile first world; we’re not bound by 40-year-old networking technology, only the laws of physics.
To users, a 100 ms delay starts to feel ‘laggy’. At 1 second, mental context shifting occurs. By 6-10s, the user has left and may never come back. Any optimization you build in at the application layer is magnified by the raw, unadulterated speed that PacketZoom unlocks.
User Experience
A great app user experience is ruined if the user has to wait too long to interact with it. Any optimization you build in at the application layer is made better by having access to raw, unadulterated speed. For users, speed is the user experience.
Engagement and Retention
Getting people to open your app is hard. Don’t make it any more difficult. Decades of research and entire industries have shown that slow loading times can have a dramatic negative impact on engagement and retention. Faster content downloads have a direct and positive impact on engagement and retention, and that’s a fact.
Among PacketZoom’s success stories are:
Measuring your Business KPIs
PacketZoom offers a built in A/B testing module that helps you evaluate the business value of our service. By default, PacketZoom A/B testing will measure speedup, throughput and connection rescue rate for content selected by the customer. Depending on your line of business, PacketZoom can measure your custom business KPIs. Since PacketZoom SDK can view all the performance of all http traffic, it could literally compare any user activity in an A/B test.
