Showing posts with label communications as a service. Show all posts
Showing posts with label communications as a service. Show all posts

Friday, February 6, 2009

New Era of Discrete Apps?

Many changes are possible as we move into an era of Web-based applications, built and accessed as Web services. On the demand side, users will be accustomed to a new way of buying and using software. On the supply side, we will see different business models. 

Especially for communications-enabled business processes, we likely will see more reliance on something we might call "discrete applications," rather than the more-monolithic approach we have seen historically, where lots of features were purchased upfront with the buying of a switch solution, for example. 

That doesn't mean every application is efficiently provided discretely. Generally speaking, large scale tends to dramatically tip the scale towards platform-based solutions. Conversely, low volume tends to tip the scale towards hosted, Web-based approaches. 

That is a pattern we have seen for services such as business phone systems, carrier switches and server farms, for example. If a provider or enterprise has high volume and lots of users, buying and owning switches and facilities tends to make more business sense than leasing or renting services or capacity.

Conversely, small entities with relatively low volume demand almost always are better off renting capabilities rather than buying and owning their own infrastructure. 

Roughly the same sort of economic logic should come into play in a new era where more applications are built and intended for use by relatively smaller number of users than in the past. The example already is seen in the broad consumer market.

Consider even broadly-purchased applications and services such as multi-channel video services. In an older paradigm, a single provider might expect 70 percent penetration. In a competitive market, even a successful provider might expect to get just 30 percent penetration. 

That changes the economics of network investment. Where once a provider might build a whole network and expect to get customers at seven out of 10 homes, now a provider has to build a network where ultimate penetration is three homes out of 10. that means shared costs must be borne by a considerly smaller number of actual paying customers. 

In the Web sphere, roughly similar sorts of phenomena are at work: in the vast majority of cases, any single application will have a smallish number of users. The opportunity for any single application will be quite small, compared to an earlier era where most people used a small number of relatively-standard applications. 

The good news is that all of the tools and infrastructure we now have will support robust business models even at relatively lower levels of end user demand. So one of the other implications is that we may be entering an era of vastly-expanded use of "discrete applications," custom built in many cases using pre-built modules or "primitives."

That in turn presupposes new ways of packaging, pricing, marketing, delivery and support, new ways of discovering needs and building solutions to match those needs. This means more discrete apps and fewer of the monolithic sort, even though some apps will continue to be relatively monolithic because they have mass usage. 

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...