Advanced Software On-Demand Bad on Functional
Streaming
Jeong Min Shim, Won Young Kim, and Wan Choi
Electronics and Telecommunications Rearch Institute (ETRI), Daejeon, South Korea
{jmshim, wykim, wchoi}@kr
Abstract. Streaming is a technology that enables either real-time or on-demand
distribution of multimedia contents over network. Recently streaming technol-
ogy has been applied onto applications, and many deployment tools for enter-someone like you 什么意思
pri applications have been developed. Software streaming is a technology to
provide software whichever urs need on-demand in real-time by using
streaming technology without downloading and installing a full package in ad-
vance before its u. Software streaming technology has many issues that are
application load time, network fault-tolerant and etc. In this paper, we discuss
issues for software streaming technology. Then, we propo a new SOD sys-
tem bad on functional streaming called Advanced Software On-Demand
(ASOD) system. Also, we prent schemes to solve issues that are application
notavailableload time and network fault-tolerant.
1 Introduction
Streaming is the process of playing application while it is still downloading [1] [2]. Streaming has been mostly found on streaming media that lets urs listen to or view the digitized contents such as sound, animation and video, as it is being downloaded.
Recently streaming technology has been applied onto applications, and many com-panies have developed deployment tools for enterpri applications such as App-Stream’s AppStream.NOW platf
orm [3], Softricity’s SoftGrid platform [4], Stream Theory’s AppExpress platform [5] and SoftOnNet’s Z!Stream [6]. Software stream-ing is a technology to provide software whichever urs need on-demand in real-time by using streaming technology without downloading and installing a full package in advance before its u.
Software streaming technology still has many issues that are application load time, network fault-tolerant and etc. To solve the problems, we will define software by differentiating environment for launching, basic function and additional functions. Basic function is first and certainly necessary contents to launch an application. Ad-ditional functions are contents for each component (a t of the menus).
bbc听力In this paper, we discuss issues for software streaming technology, and prent ad-vanced Software On-Demand (ASOD) system bad on functional streaming using basic function and additional functions. Also, we prent schemes to solve issues that are application load time and network fault-tolerant.
H. Jin, D. Reed, and W. Jiang (Eds.): NPC 2005, LNCS 3779, pp. 257–260, 2005.
258 J.M. Shim, W.Y. Kim, and W. Choi
2 Issues in Existing Software On-Demand (SOD) Service
In existing SOD system, the client requests page contents to the streaming rver (only, when the page contents are not found in the local cache) when an application tries to process a function of the application streamed to execute. To get contents, client will nd one or more request messages to the streaming rver. Operation of an application is suspended until the client receives all required contents.
We performed experiments to measure necessary contents to launch an application. Table 1 show results of Application Launching Size (ALS) for Linux Application. As shown in Table 1, in all application, a lot of contents are required when the applica-tion is launched. This fact implies that a client must nd a lot of page requests. Also, urs must wait for a long time after they request rvice.
Network is one of the important issues in SOD rvice. If a client los connection with a streaming rver, it is not able to request a page to the streaming rver. If urs try to u the function which is not stored in a local cache, the application will be destroyed or the client system may be crashed. Conquently, if network connec-tion fails, the rvice has to be stopped although urs can u a function which is stored pages in a local cache.
Table 1. Application Launching Size (ALS) versus total size of an application
Application Application Launching Size Total size
CBtracker 3.1 MB, of total size 100% 3.1 MB
Bubble Shooter 6.03 MB, of total size 99% 6.1 MB
drifterAbiword 4.6 MB, of total size 15.7% 29.4 MB
OpenOffice 98.3 MB, of total size 41.5% 236.3 MB
3 Advanced Software On-Demand Bad on Functional Streaming 3.1 Architecture of the ASOD System Bad on Functional Streaming
To provide SOD rvice bad on functional streaming, preliminary work that ana-lyzes an application is needed. First, we define new transmission unit between a client and a streaming rver, named Functional Unit (FUint). The FUnit consists of contents for one or more menus. There are two kinds of FUnit: (1) basic FUnit and (2) extra FUnit. Basic FUnit is first and certainly necessary contents to launch an application. Extra FUnit is contents for each component (a t of th
e menus). We analyze software execution and extract necessary information for functional stream-ing. We should be able to extract this information by extracting statistical data from simulations.
Figure 1 is architecture of client and streaming rver bad on functional stream-ing. A client system in SOD system consists of the following components: (1) Streaming Application (SA), (2) Event Hooker (EH), (3) Application Streaming File System (ASFS) and (4) Streaming Data Manager (SDM). SA is an application run-ning through SOD rvice. EH intercepts functionality, is lected by ur, of an application when network fault occurs. ASFS communicates with the streaming
Advanced Software On-Demand Bad on Functional Streaming 259 rver to get FUnits. SDM stores FUnits and information for FUnits streamed from the streaming rver, and maintains information for all FUnits of an application. SDM also maintains relationship between the FUnits and functionalities.
A streaming rver consists of (1) Streaming Data Package (SDP) and (2) Data Package Information (DPI). SDP is FUnits, are extracted through Software Analyzer, for applications. DPI maintains information for application packages and FUnits. DPI is ud to find out FUnit which is corresponding to function required from the client.
Fig. 1. Architecture of client and streaming rver in the ASOD system
3.2 Techniques in the ASOD System Bad on Functional Streaming
As experimental result in ction 2.1, most applications need contents, called basic FUnit, above 40% of the total software size to launch an application. An application cannot launch until basic FUnit completely are arrived. In existing SOD system, a client nds page request message of veral tens and hundreds to streaming rver to get basic FUnit. Therefore, whenever the client requests the page, the streaming rver arches it and then transmits it to the client.
In the propod system, application load time can be reduced by using basic FUnit. When a client requests a software streaming rvice to the streaming rver, a stream-ing rver nds immediately basic FUnit and FUnit information for the application to launch the application to the client without another request of the client. Accordingly, application load time must be reduced significantly.
手语翻译
The propod SOD system supports network fault-tolerant. If a client los its con-nection to the streaming rver, the client will be notified of the situation and client may continuously u the application with the functionality that it might have. If a menu clicked by a ur is not in a local cache,
process for the menu is ignored by EH. Conquently, urs can continuously u the application, although network fault occurs.
In the propod SOD system, we u a prefetching technique to reduce an applica-tion suspension time. The prefetching is a technique that nds the FUnit to be ex-pected from the streaming rver to the client without ur’s demand in advance. There are three different ways to apply prefetching: (1) prediction by producers, (2)
软件开发培训机构260 J.M. Shim, W.Y. Kim, and W. Choi
prediction by static statistics, and (3) prediction by dynamic statistics. Prediction by producers is that the order of FUnits for the application is decided by producers with-out any statistics. A streaming rver nds FUnit to a client in quence when a r-vice is begun by request of a ur. Prediction by static statistics is that content pro-viders or packers decide a transmission-priority by using statistics collected from each ur. When an application through the SOD rvice is launched, the streaming rver first nds basic FUnit, and extra FUnits are transmitted by decided priority after. After a priority is decided, it doesn’t change. Prediction by dynamic statistics is simi-lar to prediction by static statistics. But, prediction by dynamic statistics updates a priority of extra FUnits by using sta
tistics continuously collected from each ur that us an application provided through SOD rvice.
4 Conclusion
crayon是什么意思In this paper, we propod advanced SOD system bad on functional streaming for more efficient SOD rvice. In the propod SOD system, a transmission unit be-tween a client and streaming rver is FUnit. We also prented schemes to solve issues in existing SOD system. To reduce application load time, a streaming rver nds basic FUnit without to wait for request message of a client when a rvice be-gins. The propod SOD system supports network fault-tolerant that urs can con-tinuously u functions which is stored in the local cache although network fault oc-curs. Therefore, we introduce prefetching technique bad on statistical information to reduce the application suspension time significantly.
十大幼儿英语品牌
References
1.Bitpipe, “Streaming Media Services,”
2.California Software Labs (CSWL), “Basic Streaming Technology and RTSP Protocol,”混血儿英文
3.AppStream Inc., “AppStream Technology,”
4.Softricity Inc., “The SoftGrid Application Virtualization Platform,” www.softricity.
com/home/index.asp.
5.Stream Theory, “The Enterpri Software Distribution Platform,” www.streamtheory.
com.
美少女啦啦队下载
6.SoftonNet Inc., “Z!Stream Technology,”
