8.2 Infrastructure Needs Me

8.2.1 Computing Realms

Me-centric applications will be running on an aggregate computer that will consist of many distributed individual computers. In order to understand this aggregate, we need to classify these computers into realms (see Table 8.1) so we can think about the kind of software we'll need to build for them. These realms are also illustrated in Figure 8.1.

Figure 8.1. Computing Realms

Table 8.1. Computing Realms

We see computing being handled in roughly four realms, with different functionality appearing at each level:

• Personal Area Network (PAN) The device(s) that I carry with me.

• Local Area Network (LAN) Devices that are in the local physical area that I can use.

• Metropolitan Area Network (MAN) Servers that are distributed regionally to handle load balancing, proxying, caching, content delivery, authorizing, etc.

• Wide Area Network (WAN) Servers in large data centers carefully organized to handle large loads for specific applications.

The PAN includes all the devices that I carry with me. These could include a PDA and cell phone in the short term, but may end up being a richer set in the longer term as smaller, single-purpose appliances appear. These computing resources band together to serve a single person. The LAN serves devices that are in the local physical area that I can use. These may be I/O devices like large screens, speakers , and printers. They may also be computer resources, such as local data caches, media accelerators, etc. They also include connectivity resources such as access points, connectivity diversity (a combination of beacons , 802.11, cell networks, and/or wired networks). These computing resources enhance the computing for the small set of people in physical proximity.

The MAN includes servers that are distributed regionally to handle load balancing, proxying, caching, content delivery, authorizing, etc. These may be at an ISP, in an enterprise machine room, or at a wireless regional data center (serving several cellular antennas). These computer resources serve a large number of people based either on physical location (cellular network) or based on business relationship (ISP provider or coworkers in an enterprise). The WAN provides servers in large data centers carefully organized to handle large loads for specific applications. This is the way that most Internet and many wireless services are provided today. These computer resources are optimized to scale to huge populations, covering a measurable percentage of the worldwide demand for specific services.

8.2.2 Ecosystems

We postulate three ecosystems that are competing for dominance . They differ in how services are delivered to an individual.

• Spontaneous The idea here is that appliances will form spontaneous connections to perform localized services for me without requiring any global Internet access. As an example, my PDA, camera, and printer may collaborate to provide a preview, crop, transcode, frame, and print service for real estate snapshots without requiring any external connection.

• Internet The networks (both wired and wireless) are stupid and merely carry bits between my appliances and the la carte services that I use.

• Managed My wireless network and/or my portal carefully manage, unify billing, and control the quality of services I use, though they will also grudgingly allow me to get access to others through their system.

Both the managed ecosystem and Internet expect me to use WLAN (802.11) and WAN (3G) technologies ^[1] ; they differ on business model, not capability.

^[1] 3G is the third generation of mobile technologies, which includes the UMTS technologies.

There are several ways the competition between these ecosystems may turn out. One may dominate, dwarfing the other two. Or the three may coexist (as they do today). We will look at these ecosystems as being equally likely for now and, trying to hedge our bets in the technology roadmap, concentrate on those technologies that keep our options open .

For the users, it should not matter which of the three technical ecosystems they will be using. For them, it will be one unified ecosystem that will make sure that all required services will be available, no matter how they are transported to the user . ^[2]

^[2] Seamlessness of user experience is the ideal, but seams, gaps, and drop-offs between these networks are today's norm. We expect seamlessness to be a prerequisite for rapid user acceptance. Seams create friction and drag, slowing down transition to the new paradigm.