From Steeple

[edit] 1 WP/2

[edit] 1.1 Initial technical setup (11 weeks effort, during months 3-5)

• 2 Weeks for Oxford to setup new SAN storage hardware
• 2 Weeks for initial Replay system to be setup for test purposes
• 2 Weeks to document the approaches taken with both aspects.
• 2 Weeks to setup new PcP Processing node (ideally using Episode Engine)

Some consideration needs to be given here to reporting on technologies, capabilities and implications (e.g. networks/comms and network capacities; storage requirements dependant on file types and durations; deep archiving practices)

Work package 2 is initial architecture work building upon Oxford’s current implementation of Podcast Producer documented in WP 1.1 and also informed by requirements outlined in WP 1.1, WP 1.4,1.5,1.6.

The project needs to install a new server testbed and document the hardware requirements of the server, supporting server infrastructure and the storage and back-up scenarios for low and medium usage. The project will need to process and store extremely large amounts of video and audio data at an enterprise level. Therefore considerations must be made for a modular approach to the architecture built around server resilience; any developments must fit as much as possible with existing centrally supported architecture. Any online hosting requirements must be resilient and secure, and network security concerns must be explicitly stated. The online storage of material should be placed on a dedicated SMB/AFS network or similar to isolate and protect the encoding engine from public online vulnerability. This work will document a typical testbed and use it to review set encoding engine options, which may be commercial or open source (e.g. Podcast Producer or ETH Zurich's open source engine REPLAY), or a combination of commercial/open products. The project will pilot with current architecture based on Apple’s Podcast Producer and Episode encoding engine but will install, review and consider the following encoder scenarios:

• PCP and episode podcast on it's own (likely to be insufficient)
• ETH Zurich’s Replay 0.5 + ffmpeg (an open end to end solution but in early beta)
• ETH Zurich’s Replay 0.5 Replay with episode engine as transcoder
• Replay 0.5 using PCP and episode podcast as transcoder

This will cover a wide range of server encoding related topics:

1. Installation and configuration of Podcast Producer (PP) on main architecture
2. Installation and configuration of Replay on test architecture (VMware?)
3. Installing and using Episode Podcast
4. Detailed hardware requirements analysis (both for the pilot project and for any future podcasting service)
5. Integration of PP into Oxford LDAP environment
6. Initial work on Kerberos cross-realm authentication with Oxford KDC
7. Xgrid best practices

Deliverables:

• Documentation, including:
  • architecture schematics;
  • equipment list;
  • storage estimates per annum;
  • and a review of server requirements per product.
• SWOT analysis on PP and Replay.

Skills required: This is one of the more technical packages that requires insight into networking, file systems, SAN setup etc. Oxford will contribute support from across the computing services department to assist with technical expertise.