Storage in the clouds

Work Completed By: Winston Castle

Project Brief

One of the concepts within new wave infrastructure is that of cloud computing, where services such as storage or processing power are provided to a consumer without them having to worry about the underlying technical infrastructure.

This project is concerned with demonstrating how by applying the concept of cloud computing to storage a service can be provided that scales both up and down in line with demand.
Provide at least two storage ‘pools’ that are hosted within internal or external clouds and are capable of increasing and decreasing capacity in line with demand. Both must be accessible from a desktop/laptop client device.

Consider the data types and the level of security required.

If possible demonstrate how different security measures can be applied.

Management Overview

At the moment there is no understanding of Total Cost of Ownership (TCO) of networked disk space or of SAN space. We know how much it costs to buy, we have support contracts with the suppliers of the SANS and of hard disks in networked servers so if they go wrong they can be fixed. There are also the costs of the electricity to run the devices, Uninterruptible Power Supply (UPS), air conditioning, opportunity costs of the building space, making the area waterproof, council tax on them etc. as well as the support staff. Putting together these costs is a major project in itself and a necessary one so that the costs of cloud space, which are easily quantifiable from Amazon, ElasticHosts or ZumoDrive can be measured against what we do now. Otherwise the argument of cost which has to be of huge importance is merely an argument of opinion as opposed to fact. This work will try to put in context some ways of procuring public and private cloud space.


Details of where code, web pages etc can be found.


Elastic Hosts


Personal and Group Storage

Personal storage at the moment is available as the G: drive. This storage is visible only to the user and can be considered 'personal' in that sense. The storage however should be work related, not a dumping ground for holiday snaps and backups of hard disks. There are many ways of pruning data so that policies will auto delete the wrong type of files e.g. .JPG or .WMV from a disk area or auto archive data to other cheaper disk systems. Novell Storage Manager can resolve all these issue both for G: drives Personal Storage, N: drives (The auxiliary drive for the Notes implementation) and for the more 'Collaborative' areas such as H: in the WCC arena. Unfortunately whilst NSM qwas being assessed at WCC the latest version, 2.5 was not available and the earlier versions were not capable of fully controlling the WCC Infrastructure. In a work context files that are not ready for sharing e.g. work in progress can be stored here and then 'published' to the relevant part of the H: drive for workgroup access. The usage of drives should change in the long and medium terms as Sharepoint is used as a save and collaboration tool.
In the short term there is the issue of the enormous amount of personal data on the G: drives e.g. The Resources G: drive is 140 GB and the old 'Education' server has 380 GB. Part of the project to migrate to AD will have to move this data to Windows servers. The costs of moving this huge amount of data needs to be taken into account When there is only a head charge for data rather then a cost per MB there is no reason for people to limit their usage. Applying a costing basis per user and per Business Unit will help to bring these issues back under control.

The first area of Cloud Storage considered is for personal or near personal data. provided by Zumo drive. The amount of data used is directly related to cost. The first 1 GB is free then the next 10GB costs $2.99 per month. Considering that the information is 'Personal' then most users should be able to have 'free' computing here.2.99 per month. Considering that the information is 'Personal' then most users should be able to have 'free' computing here.

Security of Data on transfer and in storage

Initially there is the issue of how secure is the data on the third party area. Their answer is that the storage is secured by SSL to the Zumo servers and the backup is to three locations on Amazon's S3 datacentres using military grade AES encryption . So considerably more secure than locally held data if their website is to be believed.

Graphic Showing Security of Data in Transit and on Zumo's Servers

Security of Transportable and Shareable Data

ZumoDrive and iFolder (see below) are basically methods of distributing data. So a file on your work PC can also be on a home PC. Zumo does all the synchronising behind the scenes so emailing files to yourself becomes a thing of the past.
This ease of movement has its own problems. Do you trust users to have a copy of their 'personal' work automatically copying itself to their home PC, PDA and iPhone? A philosophical approach to this is beyond the scope of this project however there are other security issues to consider as well as this one.
A further feature of ZumoDrive also allows you to 'trust' other users via their email address. So a whole Z: drive or just a file or two could be shared allowing for small workgroups to grow organically. The updating of ZumeDrive allows another collaboration tool which is included in the Office 2007 package to be used. This is Microsoft Office 2007 OneNote which automatically saves back its data and allows for a whole gamut of useful tools from screenscraping to freehand drawing to searching as well as traditional documentation. This constant updating means collaboration of groups of users would be very easily setup when Sharepoint was not appropriate. Although this product is marketed by Microsoft as a collaboration tool often used by schools it could be used much more widely and effectively by WCC.

iFolder (WIP)
The functionality is very similar to ZumoDrive above which was probably designed with iFolder in mind.
Files are encrypted so even the admin cannot see them The only thing admin can do is view an index code which lets him/her find out who the encrypted files/folders belong to. Lost p/w dealt with by cool password reclamation. When allying this feature with the new hard disk encrypti on being rolled out at WCC the data is kept secure if kept on WCC hardware.
If passphrase is lost/forgot, iFolder client has cool data recovery for encrypted iFolders. Here the administrator has no role to play. The communication takes place between the user to the recovery agent (eg: person in CTO Office).

Server Requirements SuseLinux10.3
Client Requirements Mac 10.x, Windows XP & Vista and later, SuseLinux10.3


Describe the degree to which the work was successful in addressing the project description. Include reasons why or why not.

The project investigated three methods of cloud storage and they all had their own advantages and disadvantages

Cloud Based Personal Storage with some Workgroup functionality

ZumoDrive Personal Storage over Different types of Devices
The installation of the drive is straightforward, download some software from their website then setup an account using an email account and a password. see zumodrive.jpg
The same process can be repeated on other devices such as laptops, netbooks, iphones etc. To reduce space takeup on devices with little hard disk space files can be set to not reside on the local zumodrive and just be available as streaming files so an iPhone could be used for streaming video or .mp3 files


What immediate impact could the output of this R&D work have on the organisation – could it provide benefits without compromising our strategic approach?

The specific of this work was to see how extra storage could be added for users. The costs of storage and server hardware are unknown at the moment (see Strategic Implications for a discussion of this point) so it cannot be

Private Costs of a SAN wip

Policies were also available to spot contraband files occupying space on expensive gear so they could be brought to the attention of user managers and handled through human resources-sanctioned procedures. With contraband files consuming on average between 5 and 15 percent of storage capacity in most shops we visit, and with the costs to host these collections of bootleg music, picture, and video files topping $300 per GB when spindle costs, management costs, strain on backup processes, and many other ownership cost components are factored in (not to mention financial risks to the company from the MPAA, RIAA, and others), we could readily see how NSM would pay for itself virtually on installation just from the standpoint of file system cleanup!


How the work carried out fits with our strategic direction or how it should contribute to our strategic thinking.

The issue of Security of Cloud data is interesting. It could be that rather than worrying about losing data that is kept in the cloud. It could rather be too intransient as the following article makes clear, putting data on public data spaces means you could lose control of when and how it is deleted.

Costs of Cloud Space

A Server with a small disk

Costs include £10.00 per month for a Microsoft 2008 webserver licence


Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License