Inject… INJECT!… i-n-j-e-c-t… What is really best for in-browser notifications?


20 January 2016 | Speaker: Sergey Khromov


In the previous post (“In-browser notifications – hardcore softies”) the importance for in-browser notification systems (IBNs) to be software-only solutions was looked at, emphasizing the importance of IBN to be a software-only solution to give operators the required flexibility and security, while at the same time reducing their CAPEX. Such reduction can indeed be significant, if further enhanced and magnified by additional IBN characteristics.

One of such competitive advantages that sets apart carrier-grade and carrier-oriented IBNs from standard online advertising and notifications solutions is IBN’s ability to work with several injection methods, to best accommodate operators’ network and traffic requirements. The most popular approach today is to use proxy to processes html objects and inject messages. Several “all-in-one-box” solutions offer it and many operators today, when discussing IBNs, presume that this is a “de-facto” standard in IBNs that must be applied, despite obvious disadvantages for operators and subscribers.

While it is true that today proxy-based approach is the only way to work with https (SSL) traffic, several serious security and efficiency issues arise when this method is used to inject messages in http traffic, making it a potential single point of failure, thereby affecting the stability of the whole IBN solution, operators’ services and subscriber QoE. The last thing that any operator wants is to have its services shaken up because the magnitude of this tremor can be catastrophic to subscriber loyalty. This is why it is critical for operators to be aware that other injection methods are also available on the market and, when comparing different IBN options, understand how these methods differ and which carries most benefits, considering traffic, network and other factors. The reason is not to catalyze curiosity but to find the IBN that is best for business.

Besides proxy-based approach, there are at least two other methods that operators need to consider – “Out-of-band” and “DPI-based”. Without getting into technical details, let’s have a brief look at these approaches.

“Out-of-band” (OOB) injector relies on monitoring mirrored requests, thereby avoiding the bottle-neck issue of the proxy-based approach and at the same time significantly improving scalability and productivity of IBN because throughput (which is another serious limiting factor in proxy-based injectors) is not relevant for OOB – here the performance in transactions per second (TPS) for special HTTP requests, as well as the concurrent subscribers that can be targeted with such IBN are important and high scalability of such IBNs ensure that there are no hiccups along the way. Another key factor to be aware of – OOB approach causes no latency, as opposed to proxy-based. So if, for example, things get really bad and injector server is down, it will have absolutely no effect on subscriber experience –you will browse the web as usual but you will simply not see any in-browser notifications.

“DPI-based” injections take it even further and use DPI filtering rules to inject according to specified conditions. This approach reduces CAPEX for operators even more, because here DPI is used as an injector, thereby eliminating the need for a separate injector server. In the example in the previous post, the software-only IBN used DPI-based approach to give the operator an incredibly flexible, scalable and very efficient and productive solution that significantly enhanced monetization and future growth opportunities.

However, for an IBN to be a true subscriber engagement solution, it is not enough to be just a universal adaptor that can plug into everything and integrate with anything. It must be given intelligence. IBNs must be able to use the vast amounts of subscribers’ data that operators already have for segmented and personalized approach, enriched by knowledge subscriber needs, preferences and interests, taken from operator’s big data and derived by IBNs from subscriber behavior. This will be the focus of the next post…


back

READY TO TALK ABOUT YOUR NEXT PROJECT?

×

Hello!

See examples of Jet Toolbar messages!

Find the icon in the bottom right-hand corner of this page.

Click on it and see some examples of messages that can be created with Jet Toolbar