[LINK] Australian Government Data Centre as a Service

Bernard Robertson-Dunn brd at iimetro.com.au
Wed Mar 2 11:10:31 AEDT 2011

On 2/03/2011 9:11 AM, Tom Worthington wrote:
> I am not sure that the government will need as much space as has been
> predicted. If bespoke applications are replaced with standard web based
> ones, the processing and data storage requirements will be much smaller.
> My estimate is a 99% saving in processing and data storage. Before you
> challenge that, let me explain it is an estimate and I am an expert, so
> I do not have to back it up with any actual evidence. ;-)
I'd dispute most of that.

You made a prediction about three years that the government's data 
centre space would fall from 10,000sm to 100sm and all be put in 
containers. As it turned out, by the end of the year the government was 
reporting that it had 30,000sm of data centre space. Part of that was 
because the first estimate was not for all the agencies and the second 
was much a more complete estimate, but the fact is that data centre 
space requirements is a growth industry.

And if you are suggesting that Centrelink throws out its Model 204 based 
applications and puts in web applications or if any of the other 
agencies that use mainframe based OLTP applications replaced them with 
web applications, then I'm afraid you just don't understand enterprise 
systems and the costs involved in replacing them.

Not only that but the government's data storage requirements are growing 
at 60-80% compound per year. This is compensated for, in part, by 
increasing data storage density, but the prediction is that data storage 
(and other) infrastructure space requirements will grow, not fall. This 
has been the trend for many years and no real expert believes otherwise.

> Gershon was essentially about saving money by rationalisation, with some
> environmental and other benefits as a by-product.
Gershon was about demonstrating saving money, not about really saving 
money. For example, his recommendation about reducing the number of 
contractors has been adopted. Unfortunately, the unforseen consequence 
(unforseen by him, anyway) has been that many contractors have been 
moved into permanent positions in the APS, mostly at much higher grade 
levels than is justified (in order to retain their salary levels). So 
now there is a lump in staff numbers that is going to cause all sorts of 
HR problems down the line. And no money was saved, in fact it probably 
costs more to employ this lot that it used to - but the cost is hidden. 
And the data centre strategy was about cost avoidance, not cost savings.

>   >  ... DCaaS attempts to address the whole-of-government problem, not
> agency problems. In fact it is likely to create problems for the
> agencies. ...
> If there is less paperwork to go through to use the shared facility and
> it is cheaper, then this should not be a problem for agencies.
Paperwork is the least of their problems. And who said it would be 
cheaper? The current set of little data rooms that many of the small 
agencies use are cheap, but unreliable. Full data centre facilities 
(even Tier 1) are more expensive than data rooms. They won't be cheaper, 
but will be better controlled, managed and monitored.



Bernard Robertson-Dunn
Canberra Australia
email:	 brd at iimetro.com.au
website: www.drbrd.com

More information about the Link mailing list