Breathing New Life into Legacy Media Servers John

Breathing New Life into Legacy Media Servers John M. Morris Coordinator: Academic Technology & Web Services Drexel University Philadelphia, PA john. morris@drexel. edu 11 -Jan-06 Educause Mid-Atlantic

What You Are Going to See • • Drexel University Overview How this project came about – Project Exigency – Media file handling processes – Project Goals • • • Project Phases Phase 1 – Pre-Pilot and Pilot Major System Components Implementation Details File Naming Convention Locating Output Types of encoding that can be done How this project breathes new life into legacy media servers Project Metrics Phase 2 – More output types and a GUI interface Demonstration 11 -Jan-06 Educause Mid-Atlantic 2

Drexel University Overview • Private institution founded in 1891 • Total graduate and undergraduate 15, 000 students • 10 Colleges and 3 Schools: Arts and Sciences Biomedical Engineering Science and Health Systems Business Education Engineering Honors Information Science and Technology Media Arts and Design Nursing Professional Studies Public Health Medicine Law (beginning fall ’ 06) • Co-op School – 5 yr. program – 3 6 -month paid professional work experiences 11 -Jan-06 Educause Mid-Atlantic 3

Drexel University Overview • First US institution with microcomputer requirement - 1983 • Campus entirely wireless - 2000 • Internet 2 connector school • Single sign-on portal • Commitment to mobility, education from any place at any time, and meeting technology needs and expectations of tomorrow’s students 11 -Jan-06 Educause Mid-Atlantic 4

Can you say the word PODCAST? This project started because a faculty member wanted to do podcasting and I saw a chance to significantly reduce staff involvement in media conversion and deployment! It quickly evolved into something more! 11 -Jan-06 Educause Mid-Atlantic

Rich Media Automation Project Exigency • Significant personnel time required to handle media file encoding and deployment. • Moderate personnel time required to manage media servers and content. • Significant personnel time required to assist faculty with access management to their media files. • Over 5000 media files on 3 media servers representing >15, 000 hours of video / audio accumulated over about 5 years. • Growth in content encoding and deployment requests increasing rapidly due to new capture apps and increased interest in podcasting. 11 -Jan-06 Educause Mid-Atlantic 6

Pre-Pilot Typical Media Handling Process Mi Raw media file produced Ms Encoded file deployed 11 -Jan-06 Mi Ms Media given to support staff Ms Media moved to server Ms Instructor emailed with URL Media file encoded Mi Mi = Manually Processed by Instructor Ms = Manually Processed by Staff Mi Instructor Mi URL puts Instructor given individual copies and to URLs into instructor pastes URLs CMS Educause Mid-Atlantic 7

Rich Media Automation Project Goals • Reduce personnel handling time! • Increase usability of existing media servers. • Prepare for changes in media encoding and deployment technologies. • Make access to server based media artifacts easier and to make all media content syndicatable (RSS 2. 0). • Use “off the shelf” encoders – keep it inexpensive! • Write it in easy, manageable code Bottom line: Create a simple-to-use media object repository out of our legacy media servers that will significantly reduce staff “hands on”. 11 -Jan-06 Educause Mid-Atlantic 8

Rich Media Automation Project R & D Phases • Phase 1 (Feb 2005 – Dec 2005) – Minimize staff handling of media files. • Create “Drop Box” automation process requiring little or no oversight by my faculty & tech support staff. • Develop RSS 2. 0 syndicatable output with an HTML equivalent. – Create an “Swiss Army Knife” encoder farm capable of encoding various media formats into stream-able formats. • Phase 2 (Jan 2006 – August 2006) – Increase ability to handle additional input / output types – Increase ability to handle metadata – Implement logging / indexing of content w/ search capabilities – Create Web-based and App-based GUI application interfaces to encoding and deployment processes. • Phase 3 (Sept 2006 – Dec 2006) – Productize system and make available to other institutions. 11 -Jan-06 Educause Mid-Atlantic 9

OK, already! What’s thing look like? Phase 1 11 -Jan-06 Educause Mid-Atlantic

Major System Components (Phase 1) • “Drop Box” folder on public access machine. – – Mountable locally (domain login) Mountable remotely (VPN + domain login) Secure Permission managed (currently limited to Pilot program faculty only) • Encoding farm (multiple encoding platforms all running the same software and all watching the “Drop Box” for content) • Media Server 11 -Jan-06 Educause Mid-Atlantic 11

Media Handling Process (Phase 1) Mi Raw media file produced A Encoded file deployed Mi A Media dropped into “Drop Box” A Media moved to server A Media file encoded Mi = Manually Processed by Instructor A = Auto Processed by new process A Mi N/A Instructor emailed with URL Instructor given puts single copies and to URL into instructor pastes URLs CMS* * Only initial one 11 -Jan-06 Educause Mid-Atlantic 12

Media Server Environment (Phase 1) • Real Networks Helix Server 9. 0. 8 w/ – 4000 max concurrent connections – 10 MB max overall output bandwidth – ~4500 + video / audio files (~14, 000 hrs) • Windows Media Server (MS Adv. Server 2003) – Concurrent users only limited by CPU &I/O performance (2000 max practical) – Outbound bandwidth only limited by CPU & I/O performance (10 MB max practical) – ~100 video presentations (~300 hrs) • Sonic Foundry Media Site Live server based on Windows Media Server (based on Windows Media Server – see above) – ~500 video presentations (1000 hrs) 11 -Jan-06 Educause Mid-Atlantic 13

Some Notes! • Number of and configuration of media servers has not changed prior to nor during pilot program (except for some minor mime and mount point configuration on the Helix server). • Servers online since: – Helix - late 2000. – Windows Media – late 2000 – Media Site Live (Window Media) • #1 – early 2003 • #2 – late 2004 • Pilot project concentrated on Helix server since it contained > 90% of the content, has players on most platforms and serve almost any type of media (content – object). 11 -Jan-06 Educause Mid-Atlantic 14

Media Server Pros / Cons (1/2) Real Networks Helix Server – Version 9 – Pros • Player & Server available on most platforms • Can serve (stream) almost any kind of media incl: – Real, Windows Media, Macromedia SWF, audio MP 3, Quicktime, MP 2, H 263/H 264 (M 4 V, MP 4), HTML, XML, PDF, TXT, … • • • Mobile extensions to serve 3 GPP * E-commerce capable (add insertion, banners, …) Skin-able players Synchronized multimedia (SMIL) Not linked to Microsoft servers or IIS. Ad-hoc mount points for live production Live, archived, and On-Demand w/ security and DRM Content Push & Pull Superior low bandwidth encoding – Cons • Not FREE (at current spec – for edu’s ~$2000/yr) * Version 11 (currently being installed) 11 -Jan-06 Educause Mid-Atlantic 15

Media Server Pros / Cons (2/2) Windows Media Server – Windows Adv. Server 2003 – Pros • • FREE Easy to configure Can serve HTTP but only through IIS Live, Archived, and On-Demand w/ security and DRM Content Push & Pull Skin-able player Ubiquitous on Windows Platforms – Cons • • • 11 -Jan-06 Will only serve Window Media files. Only Intel (Windows OS) platform Mount points must be set up ahead of live production Tied to MS IIS Individual servers limited to about 2000 concurrent connections depending on system specs. Educause Mid-Atlantic 16

Command Line Media Encoders • • • Real Network – Real Producer Plus 10. 0 (rm) ($) Windows Media Encoder 9. 0 (wmv) (Free) Switch 1. 05 (wav -> mp 3) ($) Tag (add metadata to mp 3 files) ($) Text-Aloud 2. 048 (text -> mp 3) ($) – Plus AT&T voices ($) • • On 2 Technologies (various -> swf) ($$$) Konvertor – xxx 2 wav (strip audio from avi, mov) ($) pdf. Camp 2. 10 (various -> pdf) ($) Cleaner XL 1. 5 (various -> various) ($) – Will replace On 2 Technologies, Switch and WMEncoder in Phase 2 – Will be used to create mp 4 & m 4 v 11 -Jan-06 Educause Mid-Atlantic 17

How the “Drop Box” Works Concepts & Issues • File Naming controls encoding and deployment • Separation of public systems from encoding farm and media servers for security. • Multiple encoders to keep up with processing demand. • KISS (Keep It Simple $^&*@#) 11 -Jan-06 Educause Mid-Atlantic 18

Automation Architecture Overview - 1 Encoder 1 Public Server To Do Encoder 2 Done To Do Helix Server Encoder n “Course” Name To Do Html / xml Encoders watch To. Do folder and copy content to local To. Do folder for processing 11 -Jan-06 Educause Mid-Atlantic 19

Some Process Details - 1 • Faculty/Staff drop one or more media files into the drop box for processing. • Two independent “Scheduled” processes running on encoders: – Watch the remote “To. Do” folder (A) – Watch the local “To. Do” folder (B) • If Watcher (A) finds a well-formed file in the remote “To. Do” folder, it LOCKS it (adds a. lock to the end of the file name) and copies it to the encoder’s local “To. Do” folder. • By locking the file, the other encoders will not act on it. 11 -Jan-06 Educause Mid-Atlantic 20

Automation Architecture Overview - 2 Encoder 1 Public Server To Do Encoder 2 Done To Do Helix Server Encoder n “Course” Name To Do Html / xml Encoders process media file and place results onto Helix server 11 -Jan-06 Educause Mid-Atlantic 21

Some Process Details - 2 • Process (B) on each of the encoders watches its “To. Do” folder continuously. • Selects & Processes one file at a time. • Encoding based on info contained in the file name - FNC* • Output sent to Helix server – Content folder name based on info found in the file name – FNC* – An XML file is created for each FOLDER and updated automatically every time a new content file is added to that folder – An HTML equivalent is regenerated every time the XML file is updated. FNC: File Naming Convention – see later slide 11 -Jan-06 Educause Mid-Atlantic 22

Automation Architecture Overview - 3 Encoder 1 Public Server To Do Encoder 2 Done To Do Helix Server Encoder n “Course” Name To Do Html / xml Encoders clean up by moving processed file to Done 11 -Jan-06 Educause Mid-Atlantic 23

Some Process Details - 3 Encoder cleans up after itself – Copies processed files from remote “To. Do” to remote “Done” and removes locks. – Removes processed files from its local “To. Do” folder. – Takes a 2 minute breather and then looks for the next file to be processed (contained in its local “To. Do” folder. ) 11 -Jan-06 Educause Mid-Atlantic 24

So how does the system know where to put the resulting files? The File Naming Convention 11 -Jan-06 Educause Mid-Atlantic

The File Naming Convention • Three parts – The “Course Identifier” – The “Control Character” – The “Event Name” (the actual media file) • Example – EGMT 501!Lecture 1. avi • EGMT 501 -> Content Folder Name • ! -> encoding control character • Lecture 1. avi – Event Name 11 -Jan-06 Educause Mid-Atlantic 26

Locating Output • EGMT 501!Lecture 1. avi • Creates a EGMT 501 folder on the Helix server if it doesn’t already exist. • The “!” tells the encoders to create two RM files (56 k & 256 k) and strip the audio off into an MP 3 file. • Put all of the output files into the EGMT 501 folder – name them by removing the encoding control character and replacing the extension with the corresponding appropriate extension (e. g. -56 k. rm) EGMT 501 Lecture 156 k. rm • In a folder called HTTPXML, create a folder named EGMT 501 (if it doesn’t already exist) and create or update the XML & HTML files that make it possible to easily deploy the content. Note: the name of the XML and HTML files is the same name as the folder – thus making the location of the RSS syndication file (XML) or the web file (HTML) apparent from the start! 11 -Jan-06 Educause Mid-Atlantic 27

What Types of Encoding Can Be Done? • Various Video formats into RM, WMV, SWF & MP 4/M 4 V (video podcasting) files. • Audio can be stripped from Video files to produce MP 3 files (audio podcasting). • Document, Excel, Power. Point & HTML files can be converted into PDF format. • Document files can be converted into MP 3 files (Text to Voice)! • Metadata can be included in XML & HTML syndication files by adding a “description” file along with content file. 11 -Jan-06 Educause Mid-Atlantic 28

So How Does This Breathe Life Into Legacy Media Servers? • Significantly reduces staff handling of media files. • Uses media server characteristics to stream as many types of content as possible. • “Publishes” content via dynamically updated RSS 2. 0 compliant XML file (e. g. News Readers, i. Pods, i. Tunes, Firefox …) – enables Podcasting! • “Publishes” content by providing an HTML equivalent to the XML file. • Greatly simplifies access to content. • Turns legacy server into Reusable Media Object Repository. 11 -Jan-06 Educause Mid-Atlantic 29

Rich Media Automation Project Pre-Pilot Metrics Pre-Pilot (Feb 2005 – Jun 2005) – 5 active faculty – >400 artifacts dropped and processed – Three encoders running simultaneously (1. 8 GH P 4’s, XP Pro) – Informal – develop on the fly – Conversions: • • • avi -> rm, wmv, mp 3 (strip), swf mov, qt -> rm wav -> mp 3 doc, ppt, xls, txt, rtf, html -> pdf doc, txt, rtf, html -> mp 3 – Estimated staff time savings: 400 * 30 min/item = 200 hrs. 11 -Jan-06 Educause Mid-Atlantic 30

Rich Media Automation Project Pilot Project Metrics Pilot (July 2005 – Dec 2005) – – 25 active faculty >1000 artifacts dropped and processed >2500 artifacts created Three encoders running simultaneously (1. 8 GH P 4’s, XP Pro) – Formal agreement w/ faculty using system – Continue to develop on the fly – Conversions: • • • avi -> rm, wmv, mp 3 (strip), swf mov, qt -> rm wav -> mp 3 doc, ppt, xls, txt, rtf, html -> pdf doc, txt, rtf, html -> mp 3 avi, qt, mov -> mp 4, m 4 v – Estimated staff time savings: 1000 * 30 min/item = 500 hrs. 11 -Jan-06 Educause Mid-Atlantic 31

Rich Media Automation Project Phase 2 11 -Jan-06 Educause Mid-Atlantic

Rich Media Automation Project Phase 2 (Jan 2006 – Aug 2006) • Increase ability to handle additional input / output types • Increase ability to handle metadata • Implement logging / indexing of content w/ search capabilities • Create Web-based and App-based GUI application interfaces to encoding and deployment processes. 11 -Jan-06 Educause Mid-Atlantic 33

Rich Media Automation Project Phase 3 (Sept 2006 - forward) Productize system and make available to other institutions ($? ). 11 -Jan-06 Educause Mid-Atlantic 34

The Value Added Proposition Why all of this was a valuable initiative pedagogically and educationally! 11 -Jan-06 Educause Mid-Atlantic

The Value Added Proposition • Robustness and ease-of-use drives usability and success • Significant savings in terms of personnel time. • Faculty who typically would not involve themselves in “high” technology can / do adopt. • Adds content richness and engagement to web-based classes by providing a variety of media formats. • “Cool” technology engages students more! (edutainment factor) 11 -Jan-06 Educause Mid-Atlantic 36

Rich Media Automation Project Demonstration 11 -Jan-06 Educause Mid-Atlantic

Rich Media Automation Project Demos • Example of Text -> MP 3 – http: //broadcast. drexel. edu: 8080/httpx ml/techupdates. html • Examples of Mixed Content – http: //broadcast. drexel. edu: 8080/httpx ml/Turn. It. In. html Copyright – Drexel University – All Rights Reserved 11 -Jan-06 Educause Mid-Atlantic 38

What You Saw! • • Drexel University Overview How this project came about – Project Exigency – Media file handling processes – Project Goals • • • Project Phases Phase 1 – Pre-Pilot and Pilot Major System Components Implementation Details File Naming Convention Locating Output Types of encoding that can be done How this project breathes new life into legacy media servers Project Metrics Phase 2 – More output types and a GUI interface Demonstration 11 -Jan-06 Educause Mid-Atlantic 39

The End Questions? John Morris Drexel University 32 nd & Chestnut Sts. Korman 126 Philadelphia, PA 19104 john. morris@drexel. edu 215 -895 -2369 11 -Jan-06 Educause Mid-Atlantic