The Division's network storage device, DCENAS, has been unstable for the last few weeks. It failed entirely on Friday 18 February 2005 and following recovery there appears to be an underlying hardware fault that Dell is attempting to rectify. Analysis of the problem with DCENAS has been made more difficult due to recent episodes of network instability and virus attacks.
The Technical Services Unit HelpDesk tracks jobs, those that can't be dealt with summarily, using a Web-based tracking system. Requests for support come in a number of forms: email, phone and personal contact. From the table below, it can be seen for example that in May 2004 the cehelpdesk email account received almost 450 emails but there were less than 150 jobs added to the system. No statistics are kept on the number of phone calls for assistance received by helpdesk staff, nor personal requests, unless they are added to the system.
The following statistics are therefore incomplete and based on only those jobs that required managed planning and tracking to resolve.
Note that the Feb 2005 figures are not complete: the statistics for February 2005 were gathered on 21 February 2005 in the morning and except for emails the figures already exceed those of a year ago (including the number of outstanding jobs).
While by no means complete these figures demonstrate the yearly cyclical nature of the demand for TSU support: with the period around the beginning of Semester 1 the period of greatest demand. Demand tends to drop off during breaks, and taper off over the year.
The graph should not be taken as an indicator of TSU activity: break times are the busiest period for the TSU for jobs like clearing out older facilities, installing new laboratories, rolling out software and hardware updates and other activities that aren’t reflected in the helpdesk statistics (the helpdesk statistics are just the jobs other people want the TSU to do).
Response times
Last year (2004) the comedu helpdesk system recorded 1,243 jobs. Of the jobs completed (98.8%), the mean time to complete was 10.2 (calendar) days, the median less than 5 days, and the mode 1 day. 17.5% of jobs were completed in a day or less, 36% in less than 2 days, and 68% in less than a week.
Problems raised via the helpdesk
The following chart shows the breakdown of requests by category over the past four years. Further analysis of the data show trends down in the proportion of support requests for Printing, Email, Software, Systems, Hardware and Development, and up in Media, Assets, and unfortunately Other (indicating that a better categorisation of requests is needed, or that the helpdesk is doing new and different things compared with when the tracking software was first developed in 2000). ‘Other’ requests have more than doubled from 11% of requests in 2001 to 22% in 2004.
The helpdesk support application has not been used to report statistics before, and was not developed or managed with a view to reporting data in the form above. There have been no reporting metrics proposed or agreed, so these statistics should be considered with that in mind.
Three years ago the Division had around 400 computers listed in the Asset Register. That figure today is around 800, although it includes a number of machines waiting to be written off. 548 computers have been purchased from IT Loan funds since 2001 (others have been bought from Research and Consultancy funds), and best estimates put around 681 computers currently in active use in the Division (including servers, desktops, portables and lab computers). It is difficult to be accurate about the numbers since they change constantly as computers are upgraded and new ones commissioned. Old computers never seem to die: there is always someone with a use for a recently retired computer it seems. The AUVA computers are not being returned to service on desktops because of the failure rate among hard drives and power supplies, but even some of these have been pressed back into service to give some Macintosh users better access to Callista.
Overall the trend is for more computers, but with the same number of permanent Technical Service Unit staff positions supporting them. While there is a constant search for more efficient ways to manage the installed based of hardware, and casual staff are being employed to ease the load at busy times, the increasing load is having an impact on the service that can be provided.
Development support
The TSU does not have the developer expertise to help the Division to fully exploit the resources available to it as a result of the National Institute of Language Learning (NILL) infrastructure grant, in particular the Video on Demand (VoD) server and associated off-air live streaming, recording, and streaming on demand capability.
While all of this is possible to do manually, it would be much more efficient to automate the process of taking requests for recordings, transcoding recorded programs and preparing them for on-demand streaming (including generating the code to be inserted in WebCT pages), and moving the media files to the correct place on the server.
All the hardware is now available to deliver the solution, which can be automated if we had the expertise and the time to do it as it was first intended. A proposal for funding some contracted support to help the Division realise the potential of the VoD server is being prepared.
Several new variants of the MyDoom worm infected campus PCs on Thursday 17 February 2005 and the following days. The emails containing the malicious code as an attachment were particularly sophisticated and managed to fool a number of people around the campus into opening the attachments, infecting their PC and spreading the infection.
It took a few hours for the anti virus vendor used by the campus to provide an updated signature file for the eTrust antivirus software used to protect PCs here, so campus PCs were vulnerable until eTrust could strip the malicious code from the offending emails. With the variations of the original worm coming out over successive days, new signature files were being issued daily by the vendors in an attempt to keep up.
Staff (and students) need to be more vigilant about emails and seek reassurance from colleagues or the helpdesk before opening attachments to ANY uninvited emails.
There was no impact on Macintosh users, except having to deal with the slowdown of the campus email servers under the burden of the worm, and receipt of many emails with the malicious code attached. The comedu helpdesk staff lost a day or two dealing with the impact of the attack at the helpdesk's busiest time of the year.
Several years ago the University installed two industry-standard professional video conferencing suites based on the H.323 standard for videoconferencing. The Division's Executive held one of its meeting in the facility as an orientation on 30 May 2003.
One of the facilities was incorporated by the Court of the Future project in BLIS. The other does not appear to be in use, and there seems to be no-one who is prepared to claim management of the use of the facility (it appears ICT Services is responsible for technical support).
In late October 2004 I received a request from China asking for support for some testing of the videoconferencing link between Hungzhou Normal University and UC. Hungzhaou had installed a system based on UC's installation, and were keen to test out the link between the institutions. Despite a number of discussions about the service the link has apparently not yet been tested successfully.
One of the investments the Division made with the National Institute of Language Learning (NILL) infrastructure grant money was to purchase a one-year licence to Marratech eMeeting software: allowing participants from all over the world to join a video, audio, whiteboard, text and documents chat session, if they have the camera, microphone and broadband capacity to join in (the user software is free), using just a desktop computer.
With all the other work on at the moment though the server for this is yet to be commissioned.
[update 9 May 2005] BLIS has stepped in and supported the Division's videoconferencing with Hungzhaou, successfully connecting to the Chinese facility using the H.323 standard over ISDN. With BLIS' continued support, the use of the technology for teaching internationally can be more thoroughly explored.
The experimental multicasting of live-to-air television services around the University's computer network has highlighted shortcomings in the network. Over summer there were a few issues raised by the University's Network Manager relating to "the more simple minded devices on the campus network, eg building management controllers, CarDAX controllers, some printers, in that they get flooded and hence DOSed by the stream", but by and large the system operated effectively multicasting all five local terrestrial television services around the network throughout the summer at full digital quality. The University Network Manager is supportive of the project because it tests the network and demonstrates shortcomings that should be fixed.
With the increase in network activity as more and more people return to work after the summer break, more issues are arising (especially with the use of the network to upgrade computers for the new semester) that from time to time interrupt normal use of the network.
The network SHOULD be capable of dealing with the traffic generated by the live streaming services, but some legacy, feral or badly-designed systems interfere with the wider network causing slow-downs and sometimes outright failure. These are underlying issues that have troubled us in the past and that are just being highlighted by the streaming television services. The issues need to fixed whether or not the television services continue: it is in fact our intention to add at least ten satellite services to the current five terrestrial services, for a total of fifteen or more services multicasting simultaneously around the network, potentially increasing the current activity three-fold. Once on-demand streaming is added, should the on-demand streaming be popular the additional network traffic will test the network further.
At this stage the multicasting will continue, but there may be pressure to scale back or cease the streaming to relieve the network. The Division should resist such pressure, insisting instead that the network issues be identified and corrected.
