[help]dissertation

Hi everyone, I've got a huge favour to ask.

As you may know, I'm coming into the final weeks of my BSc in Computer Science and Engineering. As a major part (3/8) of this year's grade, I've had to undertake a project.

My choice of project was secure grid computing.

Anyway, I'll cut to the chase - I've just finished my report, and would like lots of feedback on it. For those who are willing, here's the abstract:

On Secure Grid Computing In a Large Enterprise Corporation

By

Steven Marshall

Many security models for use on computational grids have been presented over recent years, but none seems to meet the needs of the Large Enterprise (LE) and consumer sectors. In an attempt to more fully understand these needs, as well as past and present solutions, this paper will present a detailed synopsis and critique of these solutions.

Further to this, we will present a modification of Role-Based Access Controls (RBAC) to both “grid-enable” RBAC and to fulfil some greater subset of the requirements of consumer and LE grid security needs than has previously been catered by other security models. We also describe a prototype implementation of this approach, with preliminary findings from performance tests.

Anyhow, if you feel like helping me out, or are just really, really bored, please download and read my paper.

Before I provide a link, a word of warning - the .doc file is 3.8 Meg, and some 323 pages (of which, 200+ are appendices - only about 75 are my "report"). This report will eat your machine, if you're not careful.

I don't ask that you read all 323 pages, just that you read the main body of the report (or some portion therein), and provide a thoughtful critique of my work here (or via PM). I'm not asking for anything major, just "this bit might sound better here" type things.

Also, I need all replies before about 5pm GMT on Tuesday, so that I can act on them.

Oh, and don't just fill this thread with off-topic chat and "OMFG d00d, thsi r0x0rs" type comments. By all means, tell me if you think what I've done is good, but don't just post mindless rubbish.

"On Secure Grid Computing..." (.doc, 3.8Mb)

"On Secure Grid Computing..." (.pdf, 1.18Mb, Kindly provided by Seph)

@Mods: I'm pinning this for the moment. Yes, I know it's abuse of power, but I think you can all understand why I'm doing it...

Thanks in advance, everyone.

I know lots of you would like an RTF/PDF version, but neither of these formats like me. I tried making RTF, and it just made a mess. I got Seph to try making a PDF, and that didn't work well either.

Please don't post in here unless you've read my dissertation.

I appreciate any support, but this thread should only be for things directly related to helping me make my dissertation better.2>

#150377

I'm at page 40 now and here are some suggestions (most of them are about style and things):

- Try to not start paragraphs with questions. (ignore that one, you said it was OK)

- When using examples, like project Zebra and Joan, go for names that have a connection with your text somehow. E.g., when talking about Internet-protocols, you can use the name Sam and Robby as names, Sam = Sender, Robby = Reiver. If something like this example is not possible, you should still connect the names to the rest of the text.

- Maybe you should have a little diagram to compare MAC, DAC, RBAC, SPKI, etc... Makes it easier for the reader to read the text and to mind-map things.

- A list of used abbreviations and terms would be handy (glossary)

- (allright, *****ing-mode, but you are mixing "we" and "I")

So far, it is looking good. The text is of high-quality and good readable

I'll continue to read now

#150394

Originally posted by AndreasV@Apr 25 2004, 08:14 PM

I'm at page 22 now and here are some suggestions (most of them are about style and things):

- Try to not start paragraphs with questions. (ignore that one, you said it was OK)

- When using examples, like project Zebra and Joan, go for names that have a connection with your text somehow. E.g., when talking about Internet-protocols, you can use the name Sam and Robby as names, Sam = Sender, Robby = Reiver. If something like this example is not possible, you should still connect the names to the rest of the text.

- Maybe you should have a little diagram to compare MAC, DAC, RBAC, SPKI, etc... Makes it easier for the reader to read the text and to mind-map things.

So far, it is looking good. The text is of high-quality and good readable

I'll continue to read now

Thanks Ave.

As you noted, the first, whilst not technically good form, is something I do in my writing, and my tutor says that's fine by him.

The second point, I could do that, but, for the what it would add overall, I'm not sure it is worth it, in the long run.

Finally, the diagrams would be very difficult, as they're very drastically different. I've tried as far as is possible to include as many diagrams as I can, but to have comparison diags wouldn't be feasible due to the massive differences.

With regards to your suggestion of a glossary, yeah, I intended to add one, but forgot. I'll get started on that tomorrow, I think, as I've been writing for the last 14 hours or so...

#150395

Ok.....first off I'm not gonna read your report as I have neither the time nor the inclination to do so.

What I am gonna do, is help you make it a PDF file..

Download Win2PDF and install it.

Once that's done, it will install a printer driver under the name Win2PDF. Now go into Word and choose Print from the File menu. In the Print dialog box choose Win2PDF as the printer and click Print.

It will now popup a dialog box asking where you want to save the PDF. Choose a location and click Save. That should do it.

#150396

Originally posted by judge@Apr 26 2004, 02:28 PM

OK. Its an interesting read and touches on several areas that I have consulted on in the past - in particular security models.

I haven't read all of it and I didn't read it in great detail - perhaps partly because I'm already familiar with much of what you are describing. However, I personally found it to be a little unstructured.

For example, you discuss various security models (and implementations of those models) and rate their suitability to the problem at hand. However you have not previously defined what criteria you are using to assess their suitability. e.g. what are the features that would make a good GRID security model and what are the features that would make a bad GRID security model. For example you state:

Without having first discussed the pro's and con's of a distributed security mode v. a centralized one:

• What are the key attributes of a grid computing environment. e.g. resource balancing, intermittent connectivity, bandwidth v. CPU, multiple machine architectures, multiple OSes etc.
  
  
  
  
  
• What are the requirements of a security model? e.g. integration with native access control, handling of clashing identities, establishment of trust relationships.
  
  
  
  
  
• What needs to be controlled? e.g. Only execute a process on host X if it is between the times of 10:00PM EST and 6:00AM EST.
  [*}How much of such decisions need to be made on the host itself, versus what decisions are made on the ditribution node?
  
  
  
  
  
• Are all of these requirements relevant to enterprise grids rather than internet grids? Careful: my laptop, for example, is intermittently connected, sometimes over a slow link. Some hosts have to be completely out of bounds to group A because they are owned by group B.
  

As I said earlier, though, these are my personal preferences only and you should present the document in a form that you think is right.

All the best.

Thanks for the input, Judge.

With regards to the bullet points, the bulk of those are far out of the scope of my project - they are much more related to scheduling/economic use of resources than they are to the "raw" security that I'm focussing on.

As for defining the features that would make a good security model, I see what you mean. I had, however, assumed that, from the comments I make, these requirements are implicit.

Finally, the unstructured-ness of my report is very much my personal writing style.

#150768

I just DL'd it. I'll finish going through it sometime this evening (now it's 9:30 a.m. here), so keep your IM on later.

#151676