MIT’s Case Study

SZABIST1

Synchronization Issues in Mobile Databases

Submitted by:G. Habib Uddin Khan

SZABISTMr. Kamran Abid

PTCL

12th National Research Conference

Theme

SZABIST2

Use of mobile technologies to mature the CRM ofMicro Innovations and Technologies Pvt. Ltd.

(Pakistan’s largest IT vendor).

Agenda

SZABIST3

Introduction Mobile Databases Business Case

Implementation Hurdles Data Management Data Synchronization

Technologies Streaming Disconnected Mode Computing Databases

Finding & Evaluations Conclusion References

Mobile Databases

SZABIST4

What Database : A structured way to organize information Mobile Database : A database linked/connected by a mobile computing

device over a mobile network (number of patients, sale, batsman/bowler stats)

Why DB Applications have to fetch information from an information

pool/repository & perform functions even when disconnected/out of range which is why there would be a large scale usage of Mobile DB Technology.

Business Case

SZABIST5MIT: Workorder Processing [11]

Implementation Hurdles

SZABIST6

Data Management

Where to keep the records?

How to deal with versioning issues?

Data Synchronization

How to manage data transfer b/w servers and mobile devices?

What would be the connectivity mechanism?

Technologies

SZABIST7

Streaming [3]

SZABIST8

What Relatively old concept Data Dumping to/from servers No persistent storage on mobile devices Server-Side data manipulation only Mobile Devices as Thin Client

Why Less computation power of mobile devices Mobile devices cannot afford intelligent applications Less storage capacity

Disconnected Mode Computation

SZABIST9

Driving Forces [3] Advancements in memory technologies. High power mobile processors Fast RAM Flash Storage

Implications [8] Easy roaming Episodic data transfer sessions Preserving large volumes

Flat / Semi-Structured Databases

SZABIST10

Challenges Where to store transactional data during disconnected time? Is there any mechanism to save data objects on a mobile device for

a time being? Problem adopting the flat file system: How to preserve computation

power consumed just for the sake of parsing flat and semi-structured data files?

Solution Custom defined flat files (Delimiter) Application logic to parse such files XML, to provide a format for describing structured data. This proved

helpful in facilitating more precise declarations of content.

Embedded Databases [2]

SZABIST11

Shortcomings of Previous Compatibility issues due non standardization of flat files Compromising efficiency due to wastage of computational resources Huge overhead in error detection

Features [2] Full fledge DBMS FUEL (Flexible, Ultra-fast, Efficient, Lightweight) Resides in appliances, e.g., smart-cards, telephones, personal digital

assistants Consumes less memory resources Examples: Personal Folders or Phonebook on chips. Vendors: Oracle Lite server, Sybase UltraLite, SQL CE etc

Embedded Databases

SZABIST12

Installation, maintenance and configuration of databases have historically been complex tasks, often requiring expert knowledge of database design and application behavior. In an embedded environment, it is not possible to require such expertise or to perform ongoing database maintenance.

Many applications perform simple to moderate manipulation of data. They need a way of storing, retrieving and manipulating the data within the application. Embedded systems are often dedicated to a single application or small set of tasks, once an embedded system is deployed, it must continue to function without interruption and without administrator intervention.

The key characteristics of these databases are that their functionality must be completely transparent to users, no explicit database operations or database maintenance is ever performed, the database may crash at any time and recover instantly. Such as Sleepy cat Berkeley DB used with LDAP

Data Synchronization

SZABIST13

Technologies are designed to synchronize a single set of data between two or more devices

Automatically copying changes back and forth.

For example, a user's contact list on one mobile device can be synchronized with other mobile devices or computers.

Data synchronization can be local synchronization where the device and computer are side-by-side

Remote synchronization when a user is mobile and the data is synchronized over a mobile network. [http://www.webopedia.com]

Data Synchronization (cont)

SZABIST14

Things to remember

Exact schema of server at the device

Introducing Audit Attributes

Metadata Table

Inline queries

Sync. Logic always outside the device application for scalability.

Synchronization Algorithm [6]

SZABIST15

Checkout Distribution of same data to all mobile users. Chargeon Collect data from all mobile users and performs manipulation

if needed (pre-processing step for final commit on server.) Checkin Updates the master record

Synchronization Techniques

SZABIST16

Interfaces (USB, parallel or serial ports) on the servers to connect through a wired medium.

A synchronization

application is invoked to establish connection

The number of users is dependent upon available ports

The data is fetched from the device using the application embedded business logic

Sync. Techniques

Wired Wireless [6] Serverless [7]

Mobile user remain connected to the network, irrespective to the technology used.

Mobiles nodes operate in an Ad-Hoc Network. (Clique)

Efficiency of network is dependent upon the number of nodes.

Large number of connections may result in conflicts (ID address space, versioning)

Web servers, accessed by mobile users using data carriers (WiFi, GPRS, CDMA)

Business Logic is encapsulated in the form of web service

Several Concurrent transparent connections

Findings & Evaluations

SZABIST17

Generations

SZABIST18

Gen 1 Gen 2 Gen 3 Gen 4

Architecture Streaming DC Computing DC Computing Peer 2 Peer

Technology Data dumping Wired Sync. Wireless Sync. Server less

Management Only on ServersFlat Filing & Semi -Structured DBs + Servers

Embedded DBs + Global Services + Multiple Servers

Embedded DBs + Mobile Apps

ResourcesRAM (can hold only transient data)

Flash Memory Memory on Smart Cards

WLAN on Smart Cards

Usage Low Mobility High Mobility High Mobility 100% Mobility

Graphical Representation

SZABIST19Gen1 Gen2 Gen3 Gen4

Usage

Resources

Management

Technology

Architecture Streaming

DC Computing DC Computing DC Computing

Data Dumping

ServerlessWired Sync.Wireless Sync.

Semi Structured

Embedded DB’s + Global Services +Multiple Severs

Embedded DB’s + Mobile AppsOnly on Servers

RAMFlash Memory

Memory on Smart Cards WLAN on Smart Cards

Low mobilityHigh Mobility

High Mobility 100% Mobility

Solution

SZABIST20

At a glance, generation 2 and generation 3 fit into our requirement domain.

The wired and wireless technique is purely dependent upon the market client whether they need any data carrier medium or they just sync the devices by physically connecting them to the servers.

In our case, once if the Workorder is assigned to the FT, it will be locked for all others; so the problem of same data availability has also been solved.

Our case deals with the customer centric CRM and even if we follow the wireless synchronization approach the customer will suffer in case of workorder rejection by any FT. The solution to this is the implementation of Serverless environment, in which FT’s will remain updated on the move.

To make efficient use of computation power, this CRM must have to prefer Embedded databases.

Embedded database vendor may be dependent upon the technology used at the server-side to provide better handshaking.

Conclusion

SZABIST21

Data synchronization from mobile databases to the server is an issue that still persists with the researchers.

New technologies have come by to increase mobility for remote users.

In our case, mobility was the need of the hour. Since our main purpose is customer facilitation and CRM is there to fulfill this very purpose.

Adding mobility would result in an increased customer satisfaction experience, since we can provide the customers faster services. We found that wired and wireless synchronization with embedded databases works well with our requirements.

Server less is a relatively new technology and has some issues which need to be addressed. In future, with the maturity of newer technologies like server less, we can further enhance our project to cope up with the upcoming standards

References [1] Weider D. & Yu Sunita Sharma, “Mobile Database Design Methodologies

for Mobile Software Solutions” Computer Engineering Department, San Jose State University, San Jose (Silicon Valley), CA 95192-0180, USA

[2] Action members 1EPFL, 2U. Grenoble, 3INRIA-Nancy, 4INT-Evry, 5U. Montpellier 2, 6U. Paris 6, 7U. Versailles “Mobile Databases: a Selection of Open Issues and Research Directions”

[3] Anil Nori, Microsoft Corporation, “Mobile and Embedded Databases” SIGMOD’07, June 12–14, 2007, Beijing, China. ACM 978-1-59593-686-8/07/0006.

[4] Gary D. Walborn, Panos K. Chrysanthis “Pro-Motion: Management of Mobile Transactions”

[5] Margo I. Seltzer, Michael A. Olson “Challenges in Embedded Database System Administration”

SZABIST22

References (cont) [6] Mi-Seon Choi, Young-Kuk Kim, Juno Chang, Proceedings of the 2005 International

Workshop on Data Engineering Issues in E-Commerce. "Transaction-Centric Split Synchronization Mechanism for Mobile E-Business Applications"

[7] Henry Larkin, 6th IEEE/ACIS International Conference on Computer and Information Science, "Applying Concurrent Versioning to Serverless Mobile Device Synchronization"

[8] Mariano Cravo Teixeira Neto and Ana Carolina Salgado, "Hoarding and pre-fetching for mobile databases"

[9] Takahiro Hara, 16th International Workshop on Database and Expert Systems Applications, "Data Replication Issues in Mobile Ad Hoc Networks"

[10] Patricia Seybold Group, “An Executive’s Guide to CRM" How to Evaluate CRM Alternatives by Functionality, Architecture, & Analytics

[11] Technical Documentation, Fleet Manager.

SZABIST23

Q & A

SZABIST24