
The AdTech Industry faces two challenges.
A strict time constraint on the RTB bidding process and high transit costs.
During the RTB bidding process DSPs have to respond to bids within 100ms. The process of ad delivery has to be completed during the time it takes for a web page to load.
The RTB protocol is a text based protocol that can be heavy without compression. All RTB communications take place over the Hyper Text Transfer Protocol (http). The Http protocol adds to the weight of the RTB protocol.
The AdTech industry encompasses technologies, software and services used for delivering, controlling and targeting online ads. The AdTech industry has seen phenomenal growth over the last decade and a half.
This growth is driven by Programmatic and Real time bidding (RTB). Both technologies will account for $20 billion by the end of 2016. There are a couple of important distinctions between the two.
Programmatic ad buying is the practice of automating the buying and selling of digital advertising.
Real time bidding (RTB) is the buying and selling of online ad impressions, through real-time auctions. These auctions occur in the time it takes a webpage to load.
These bidding techniques result in many advantages for publishers and advertisers. At the same time, they contributed to making the AdTech industry more complex. It has increased the number of Adtech players along with the need for interconnections among them. RTB has also placed a strict time constraint on the ad delivery process.
Time Constraint on the RTB Process
During the RTB process, bids from DSPs have to arrive at the ad exchanges or SSPs between 80 to 120 milliseconds (ms). This time constraint varies depending on the ad exchange or SSP.
The time limit comprises the time required to compute an optimized bid. It also includes the network transit time.
DSPs usually allocate 40 ms to computation time and 20 ms to network congestion. This only leaves 40 ms for round trip network transit.
High Transit Costs
Each ad delivered during the RTB process requires many back and forth requests. These requests are between the various AdTech companies and the user browser.
The communication happens over the http protocol which is a text based protocol. The RTB protocol itself is a text based protocol. Both protocols add to the weight of the requests, and can lead to higher network traffic.
The AdTech industry exchanges approximately 465 terabytes of http request data each day. Most of this data is exchanged over non-optimized network routes. This leaves a lot of room for cost reduction.
Minimizing Network Latency and Transit costs
Datapath.io helps DSPs, SSPs and ad exchanges reduce network latency and transit cost. To understand this, we first have to look at Internet traffic routing.
Internet Routing Policies
At a basic level, Internet routing is governed by BGP (Border Gateway protocol). BGP decides which route Internet traffic takes through a network. In most cases, it sends Internet traffic over the shortest path to its destination. This shortest path is not always the fastest, nor is it the least expensive.
As Internet traffic is all routed through the shortest path, these paths can become congested. This leads to delays in the delivery of data packets. Also, BGP does not consider transit costs when making make routing decisions. Transit costs can vary between different transit providers.
Re-implementing BGP (Border Gateway Protocol)
Datapath.io re-implements BGP to make routing decisions based on network latency and transit cost. AdTech companies can define rules to keep latency below a certain level. Once a latency rule is defined, Datapath.io scans all network routes to find one that meets the network latency requirements.
Once these paths have been identified they are further analyzed to determine the ones with the lowest transit cost. Internet traffic is then routed through network paths that meet the requirements.
The image above is a snapshot of Datapath.io’s customer backend. It shows the network latency encountered through the default Amazon Web Services (AWS) route. In the graph it is compared to Datapath.io’s optimized network routes. There is an improvement of 20% in network latency from Frankfurt to Amsterdam.
This time can be allocated to computation time to come up with optimized bids. Quicker responses also ensure that DSPs win more bids. They also get access to sales channels with premium ad inventory that need fast response times.
Using the Datapath.io low latency platform results in significant cost savings. This occurs by eliminating the need to establish many servers near different SSPs and ad exchanges.
Datapath.io also allows AdTech companies to choose network routes with the lowest transit cost. This can occur within the latency constraint of the RTB process. Once a network latency rule is defined, Datapath.io automatically scans different transit providers for the ones with the lowest cost. Datapath.io then uses those to route Internet traffic. This results in a reduction to transit costs for AdTech companies.