Abstract
Recent advances in device technology and connectivity have paved the way for next generation applications
that are data-driven, where data can reside anywhere, can be accessed at any time, from any
client. Also, advances in memory technology are increasing the capacities of RAM and Flash, and their
costs down. These trends lead to applications that are mobile, embedded, and data-centric. This paper
presents an overview of mobile and embedded database systems and applications.
1 The New Environment - Mobile and Embedded
Recent advances in processors, memory, storage, and connectivity have paved the way for next generation
applications that are data-driven, whose data can reside anywhere (i.e. on the server, desktop, devices, embedded
in applications) and that support access from anywhere (i.e. local, remote, over the network, in connected
and disconnected fashion). Memory sizes have gone up and prices have come down significantly; with 64 bit
addressability, it is not uncommon to configure servers with 8 – 16GB of memory, and desktops with 2 – 4GBs
of memory. With advances in flash memory technology, large flash drives are available at reasonable prices.
Computers with 32 GB flash drives are making way into the market. Flash drives not only eliminate seek time
and rotational latency they consume significantly less power than conventional disk drives, making them ideal
for portable devices. All this naturally leads to mobile, disconnected, and specialized application architectures.
Application components can run on different tiers, in different service (autonomy) boundaries, and on different
platforms (e.g. server, desktop, mobile). These new breeds of applications fall into one or more of the following
categories:
1.1 Mobile
As more users adopt Wi-Fi enabled laptops, and with increasingly capable mobile devices, the need for mobile
applications is increasing. Applications like Email, Calendaring, CRM (Customer Relationship Management)
are already targeting mobile devices. Middleware infrastructures like application servers and workflow services
are becoming mobile-aware. Some reasons for such mobility trends are:
• More employees are mobile. Email and offline access is becoming pervasive.
• Mobile usage is broadening. Mobile usage is already prevalent in certain vertical domains like Healthcare,
Insurance, and Field Services.
Mobile applications are more than just browser windows – more and more applications now run natively
on mobile devices.
Data management and access on mobile devices is central to mobile applications. As mobile applications achieve
widespread adoption in the enterprise, mobile and embedded DBMSs – needed to support such applications
become an important part of the IT infrastructure. And as these applications grow more disconnected and
sophisticated with increasing data sizes the need for rich data processing capabilities increases.
We also note that consumer and home user applications are becoming web-centric and data is being stored
on the web ( cloud) and accessed from anywhere, at any time. Even data that was traditionally stored in PCs
is migrating to the web (cloud), thereby unlocking the data access from a specific location. Mobile devices
complete the anywhere data access vision – they provide access from anywhere. Smart mobile devices combine
multiple functions of phones, media players, PCs, etc. Such devices are becoming powerful in their processing
power and provide larger storage capacity. These advances provide data access and also enable caching of data
(from original sources) that can be processed offline
1.2 Streaming
In certain scenarios, data is simply streamed intelligently through application logic. There is no additional need
to store the data – except transiently for the duration of the operation. Conventional database systems require
data to be first loaded into the database; then the operation is performed, and the data may be later removed from
the database. This adds significant complexity to the application, and dramatically reduces its performance and
throughput.
Consider, for example, RFID data in applications. RFID tags are portable sensors that communicate over
specialized protocols with RFID reader devices. An increasing number of applications have begun to utilize
RFID technology including Manufacturing (e.g. Vendor Managed Inventory, Assembly scheduling), Distribution
(e.g. Goods delivery, Shipping verification), Retail (e.g. Shelf stocking, Receiving, Store Replenishment
and Ordering, Theft, Merchandise Location), Healthcare Industry to identify patients, Tracking books and automated
checkout in libraries, etc. In all these applications, RFID readers (devices) generate events when they
identify RFID tags. The RFID event streams are filtered, aggregated, transformed, and correlated so that the
events can be monitored in real-time. For example, when a truck carrying packages of product parts (with RFID
tags) enters (or leaves) a warehouse, the readers generate events. Spurious events are filtered; related products
are aggregated; the event data is transformed and presented on a monitoring dashboard in real-time. The event
processing is data-centric and typically requires an in-memory rules engine and query processing.

Download full report
http://citeseerx.ist.psu.edu/viewdoc/dow...1&type=pdf