CDT Cluster Softwre

This is the final report for DevProj:332

Description

This project delivered the full required software requirements for the cdt cluster.

Operating system.

From the outset the users wanted to keep their options open and there was the possibility that students might be given bare metal access to the nodes. the preferred options being in order:
  • DICE
  • other os on virtualbox
  • other os on KVM
  • Bare metal
As the project started in line with the development of DICE SL7 on desktops we gave the cluster purchasers the option of

General survey design

The survey was designed to try to obtain the maximum amount of information from the minimum of input, by using default values and allowing sensible multiple responses. Having a large number of comments fields hopefully meant that if we were providing completely the wrong answers people could tick "other" and explain what they were doing. Even a complete null response would generate meaningful information although we didn't get any of those

Target audience

The survey was targeted at all staff and research students in order to cover people in the school who:
  • Specify services for teaching and reserach.
  • Are most likely to handle sensitive data
  • Are most likely to be responsible for I.P.

In general we got a reasonable response although the return from admin staff was dissapointing

Time taken

Amount: 5 weeks Some of this included learning to use webmark (probably 3-4 days) and there was a further 3-4 days spend playing with various bits of software in order to decide how to generate the graphs for the report.

Conclusions

Webmark is fairly good at generating this kind of survey, the only real let down was that it wasn't possible to generate graphs straight from the returns but this would have been beyond the scope of the software as originally specced so it's an unreasonable request.

oocalc is useful for quickly throwing together results from csv files but rapidly becomes unmanagable as the complexity of the data and the questions you would like to ask of it increases. generating results of conditional queries involving multiple responses (i.e. queries along the lines of "Of the people who are currently using VMs how many are interested in a KVM service and how many of them would be willing to pay") rapidly becomes unmanagable and fraught with the possibility of error. In retrospect dumping the results into a database and using python or R would have been a better approach, albeit with a higher learning curve.

We should have had some of the anaysis software in place during the trial because we missed out on some information relating to multiple questions because people responded in ways we didn't quite expect.

More by luck than judgement we seemed to hit a sweet spot at the time we sent the survey out (8.30), most of the responses came back fairly quickly and we seem to have caught people when they were responsive before they'd settled down to work for the day.

-- IainRae - 04 Aug 2014

-- IainRae - 11 Oct 2017
Edit | Attach | Print version | History: r7 | r4 < r3 < r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r2 - 12 Oct 2017 - 09:01:58 - IainRae
 
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
This Wiki uses Cookies