First, to define 'Real-time'
Real-time is no CDN or Cache latency. When there is new data in the database, it's available to the end-user.
Real-time is not needing to hit the refresh button to see new information. It's when information folds into the page while you're reading it.
Real-time is a new volume and velocity of data. A lot of web data used to consist of 'Blog Posts' or 'News Articles'. Documents. Real-time web data is about activities. Granular, human readable micro-stories about the activities that users make.
"I read this", "I rated this", "I commented on this", "I shared this", "I edited this" and so on. Why? Because capturing, surfacing and socializing real-time activity data is part of the core essence of the social web. The ability to see not just the result of actions by users, but the play-by-play stream of those actions along side faces, names and time/date stamps takes an experience from a static 'snapshot' into a living, breathing stream. Further, by enabling users to like, reply, flag, share and otherwise interact with these activities, sites are creating new opportunities for engagement, conversation and conversion.
Real-time is a presentation metaphor. It often (but not always) takes the form of a reverse chronological stream with nested comments and likes. It helps users understand the order of things and mixes content with conversation in a way that drives engagement and return visits.
Real-time means filters instead of facts. Let the user decide what they want to see - to craft an experience that makes sense for them, and their friends.
Now, what is 'Real-time as a Service'?
If all the things above are true, then it changes everything we used to know about web infrastructure, databases, user interfaces and tools for moderation or curation.
APIs can no longer be request-response. Databases must now store far more data at far faster rates. User interfaces need to factor in names, faces and actions. Moderation and curation tools must leverage algorithms, crowd sourcing and real-time flows.
Real-time as a service, then, is cloud infrastructure that helps make this transition easier.
It is a database that can handle new magnitudes of scale - handling hundreds or thousands of write events per section. Not just to a flat table, but to a hierarchical tree of arbitrary activities.
Site -> Section -> Article -> Rating -> Comment -> Reply -> Like.
It's a database that can store all items permanently so that users can visit old streams at any time. Permanent storage that can also handle localized annotations. Localized annotations are the ability to modify the metadata of an activity - say a Tweet (Promote it, tag it, retarget it in the tree etc) - in such a way that that your view of a tweet is different from another customer's view.
It's a database that enables not just the ability to perform an SQL-like search query, but also continuously updates you when the data changes - so that you can modify the UI on the fly.
It's a database that returns not just flat query results, but a hierarchical tree - allowing you to present the activity in context.
It's a database that handles not just a few hundred users requesting (reading) data, but a few million users swarming to see the latest action in a sports game or a concert.
It's a database that organically makes connections between items by understanding the relationships of URLs and #tags to make implicit links in the graph where and when they're needed. For example a tweet mentioning acme.com should be attached to Acme.com in the tree.
And most importantly, it's a database company that understands that the opportunity of the Real-time, Social Web is far too big and moves far too quickly to possibly be built by a single vendor. A company that, as a result of this understanding, chooses open standards over proprietary formats; Partnership with best-of-breed partners over trying to build mediocre versions of everything by itself.
Polls, Ratings, Comments, Live Blogging, Forums, Data Bridging, Data Enriching, Visualization, Moderation, Curation, Analytics Game Mechanics, Authentication... the list is endless. They are all transformed by the Real-time web. They must all be part of Real-time as a Service.
And finally, Real-time as a Service is about service. Enterprise grade support. Best in class uptime. White label.
That's Real-time as a Service.
Further Reading