[LINK] Look to the clouds not the slates

[Steven Clark]

On 19/12/10 11:25, Tom Worthington wrote:
> According to media reports, Microsoft is to show Windows based slate
> (tablet) computers at the Consumer Electronics Show in January:
> <http://www.telegraph.co.uk/technology/microsoft/8201171/Microsoft-to-unveil-new-slates-and-Windows-8-preview-at-CES.html>.

I love the picture of Ballmer: he's about to take off, perhaps ;)

> I suggest not getting too excited by this. Microsoft have a long way
> to go to catch up to Apple with mobile devices. The Apple iPhone
> dominates the upper end of the smart phone market and Google Android
> the budget end. I expect the same will happen with tablets. Google
> are maturing their Android system for tablets.

I see devices increasingly becoming networked interfaces to computing
resources: and as far as most general users are concerned, they don't
seem to care much about *how* things get done, so long as they can get
things done when they want/need to. Herein lies the great
promise/opportunity of mobile networked devices and distributed computing.

But it will also put a lot of strain on the weak point of this scenario
- the network. The NBN can assist here, but more can and should be done
sooner to expand the scope and scale of network accessibility at the
demand end.

> In corporate terms what is happening with "Cloud" computing is
> perhaps more interesting than tablet computers.

Methinks the two are convergent/coextensive technologies. As more
computing shifts to distributed availability, portable networked devices
will become more attractive.

> If you are using web
> based cloud hosted applications, be they from Google, Microsoft,
> Amazon.com, or someone else, then you don't need much more than a web
> browser in your desktop computers or tablets. It does not matter much
> what operating system is used on the client device as long as it has
> a web browser to access the corporate application in the cloud. You
> can then use the same corporate applications on smart phones, tablet
> computers, netbooks, laptops and desktop computers. You need not
> worry about the compatibility of your corporate applications running
> on all these devices nor arranging long term retention of data on
> them.

This, to my mind, is exactly why tablets are taking off. Lightweight,
functional, and darn pretty devices that are more than enough as an
end-user interface to (a) more extensive computing infrastructure(s).
Tablets provide a screen large enough to actually do things on, to see
enough words/data/etc to be comfortable to use for general computing.

What is needed, in concert with expansion of the network backbone(s), is
greater localisation (proliferation) of wi-fi portals (and the like) to
the network. Better and more reliable coverage to enable more consistent
accessibility to 'the' network will see a further explosion in the
usability of lightweight portable devices.

Being able to move through wi-fi 'cells', seamlessly and securely
transferring between wi-fi cells, the way that mobile phones can connect
to the next tower as we move, will be crucial soon.

To my mind, we have always needed a low(er) power, low(er) cost network
of smaller wi-fi mini-towers that can be and should be spread across the
urban infrastructure. No reason why this couldn't be a combination of
relatively open, publicly accessible grids, with other private grids
within/across. For example, one or more generic grids across CBDs and
along major corridors, etc IN CONCERT WITH broader 3G/4G/etc networks
(I'm no fan of the view that you can only have one kind of 'wireless'
network in a space). With major enterprises (including government)
installing their own grids for their own (internal) traffic.

Better connectivity inside and around buildings. Network shadows in CBDs
are not just annoying, they're disabling.

> There are some news reports of the SA government considering cloud
> computing, after the success of some of the SA education sector's
> success with using a Microsoft hosted service:
> <http://www.computerworld.com.au/article/371731/south_australia_makes_moves_public_cloud/>.

The universities here (and elsewhere) have already moved their student
email off onto Google- or Microsoft-hosted 'clouds'.

The major concern I see is the security and privacy of these
arrangements. Many educators handle (and generate) records with
sensitive contents: student records, research data, administrative
materials, papers-in-progress, etc. APRAs warning hold for almost any
organisation. Loss of data is bad. Data leakage is also bad. But so is
loss of access to/availability of your data.

>Ian Foster from Argonne National Laboratory talked at ANU on using
> software as a service for large scale scientific computing recently:
> <http://blog.tomw.net.au/2010/12/software-as-service-for-science.html>.

Shared resources are pretty much the epitome of collaborative endeavours
:D And provides/creates opportunities for interactivity as well.

Not to mention reducing costs of support systems so more science can get
'done' on the generally limited budgets of most scientific enterprises.

[Will leave ranting about funding for science vs engineering which
relies upon that science to have anything to engineer *with* :p
Especially as I tend to (want to) fall into both boxes >.<]

>In my view using cloud computing for corporate services is a very high
> risk strategy, as it can result in loss of control of data. APRA has
> warned the banks about this:
> <http://blog.tomw.net.au/2010/11/banks-warned-about-cloud-computing.html>.
>
>A better approach is that suggested by Yusuf J. Mansuri, First Assistant
> Secretary, ICT Strategic & Corporate Services Division, Department of
> Human Services at "Cloud Computing Conference and Expo 2010" in
> Sydney in September, where he hinted at an Australian Government
> Cloud Computing Centre:
> <http://blog.tomw.net.au/2010/09/australian-government-cloud-computing.html>

This would be a fairly natural direction for FedGov, in light of their
continuing process(es) of data centre consolidation *and* expansion
following the Gershon Report. They already have a lot of virtualisation
going on in-house, so one could argue they are already using 'clouds'?

> The idea would be to have a shared facility on Australian territory, so
> it is subject to Australian law and shared by organisations with
> compatible procedures and aims.

Agreed :D

> Universities would similarly be a group with common aims who could
> use a joint facility. We have more than enough technology to allow
> secure sharing of such facilities. The issue is if we have governance
> and financial structures up to the challenge.

These social issues do tend to be the more contentious, and more
difficult to resolve.

And this connects with your point above about location in Australia:
being subject to the laws of the same jurisdiction offer familiarity,
reach-ability, and comprehensibility to all parties. The legal
frameworks are likely to be more familiar, the other parties can be
'reached' if litigation or ADR are required, and the whole scheme is
likely to be more comprehensible to the parties - leaving them 'just'
the actual shape of the scheme to nut out :D

> Previous attempts at
> shared IT services between universities have mostly failed as they
> depended on sharing development of new bespoke software.

Even collaborations founded upon of-the-shelf software have a rocky history.

> Sharing a
> data centre and hardware is much simpler. One obvious way would be to
> contract the service out to the private sector. But another would be
> to use a non-profit organisation like the National Computational
> Infrastructure <http://nf.nci.org.au/> and AARnet
> <http://www.aarnet.edu.au/about-us/history.aspx>.

IMNSO I would think that such a move would fit naturally into the scope
of a rejuvenated AARNet.

> Apart from saving organisations a lot of money (and CIOs a lot of
> worry), this central service approach will also save a lot of
> electrical energy and therefore carbon emissions.

Although too much centralisation runs the risk of exposure to universal
collapse should anything befall the data centre. Continuity of service
is not an inconsiderable issue for any large enterprise. Not
insurmountable, but it does lurk out there.

It's not (just) about server availability - network capacity and
accessibility is a big concern: especially for those of us in the
(mobile) casual academic pool, or working and studying  full/part-time
and thus moving on and off campus a lot.

This is one of the big areas for potential growth for a better
distributed network plus tablet/slate-type devices. (And also smaller
'smart' devices).

> However there are a lot of corporate issues to resolve in sharing a
> hosted computer system. One easy approach is to outsource the
> provision of a specific service which is hosted remotely. An example
> of this is when a educational institution changes to a new Learning
> Management System (LMS). Because the LMS provides a relatively
> self-contained set of services and uses a web interface, it can be
> easily provided by an external hosting service. The university just
> needs to provide a network for desktop and mobile web clients to
> access the remote service. They do not need to run a a money and
> energy hungry server, or maintain educational software on the
> clients.

Though this does raise issues for (staff and student) training:
especially when you no longer control 'upgrade' cycles. This is not a
trivial issue - a great many people are computer literate in the sense
that they 'know' a limited set of repetitive movements to use to get
devices to do X or Y: change the interface and they're lost at sea.

[Gen Y or whatever is no better at *understanding* technology than
anyone else without real education. Being familiar with interfaces is
not the same thing as understanding what's going on behind them, or why
they're constructed the way they are. It *does* (or at least can) reduce
barriers to engagement.]

> There are companies which specialise in providing LMS
> hosting. This allows the company to optimise their computer system to
> run just one type of software for multiple clients. Both ANU
> <http://information.anu.edu.au/daisy/infoservices/1802.html> and
> University of Canberra
> <http://www.canberra.edu.au/tlc/lms-renewal-project/lms-renewal-project>
> converted from an internal Web CT to externally hosted Moodle
> service.

As open source software, Moodle has advantages in being cheaper to
acquire and (hopefully) to maintain as well. It is also more flexible
than WebCT (I've used - and been abused by - both :) but that
flexibility can also be a burden.

> The same can be done for accounting, CRM and other software
> services.

And AFAIK, is being tried - such services are beginning to be marketed
to SMEs.

-- 
Steven