Discussion Paper on Data Centre as a Service (DCaaS)

Author: 
Kayelle Wiltshire - AGIMO
The Department of Finance Archive

The content on this page and other Finance archive pages is provided to assist research and may contain references to activities or policies that have no current application. See the full archive disclaimer.

My team is implementing the Australian Government Data Centre Strategy (the Strategy) announced on 22 March 2010. One of the elements of the Strategy is something we called ‘Data Centre as a Service’ (DCaaS). This is our shorthand for the services that will address the challenges faced by the fifty or so smaller government agencies.

The goal for DCaaS is to provide more efficient and effective solutions for data centre based ICT for the smaller agencies. The discussion paper defines the problem and offers principles and approaches. We are seeking your feedback to help us select the better approaches, identify constraints and define the requirements for any future procurement activity.

You may also have seen AGIMO release a draft paper on the Australian Government’s possible use of cloud computing. In this paper, it was noted that the Data Centre as a Service (DCaaS) project will consider cloud technologies in providing common data centre facilities and ICT solutions.

I look forward to reading your comments, suggestions, etc posted on this blog. Comments will remain open until 18 March 2011 (update) 21 March 2011. Well, that’s all from me, we look forward to hearing from you :-)

References:

Comments (9)

Kayelle

Thanks for the opportunity to comment and congratulations on your use of a blog for feedback on your DCaaS Discussion Paper

With regard to the scope, it would be useful to incorporate an implementation strategy along the lines of offerings of compliant services.

By compliant, I mean those conforming to the security, accessibility and transparency obligations on Agencies.

In this regard, I would suggest webpublishing and email services should be a quick wins. These are mature cloud services and operationally, the most vulnerable.

Authentication would be lower down, if I were preparing your strategy. Video would be an area I would prioritise.

The funding/pricing model and definition of products/services is crucial to the success here.

Marghanita da Cruz

Thank you for your feedback, Marghanita. Our research with candidate agencies is telling us that a full spectrum of service offerings is being sought. Some just want to house their ICT infrastructure and will continue to maintain it. Other agencies would like a managed service for the provision of their ICT.

Hi Everyone I have been asked a question off-line, could DCaaS be used by local governments?

As our focus has been on smaller Australian Government agencies we have not specifically addressed this in the paper. In general our work has recently allowed State and Territories to opt in. The team and I will give this some thought. My broader question to the Industry is how could/would this work? cheers Kayelle:)

We received a fairly comprehensive response to the discussion paper earlier this week. As the replies were lodged using the structured fields in the discussion paper, these cannot be seen on the blog. We are producing the original responses, and our replies, below. Please note that this is the first part of two.

Question 1 of 5: Is this scope realistic and will it achieve worthwhile results? Should the scope of DCaaS include applications, and if so, which ones?:
While the centralising and sharing of applications like e-mail, and services like storage is a good idea, the reality of such a scheme is somewhat messy.
Take for example e-mail and the controls around its use, will cause much angst among potential users, currently a Department has virtually autonomous control over such things as, configuration, individual storage limits, exceptions allowed within a Department, etc. Removing this ability, which I suspect a DCaaS supplied service will require, will not be popular and will be resisted by many.
Perhaps the best way to implement this would be to standardise the use, and configuration of e-mail first, ie forcing compliance with a configuration irrespective of joining a DCaaS supplied service. Then opting into the service would not be seen as a loss of control.
The other major issue with centralising and sharing services, is scheduled down time. You will never be able to schedule an outage that will appease everyone. Agencies that have no operational roles, being forced into an outage after-hours is not a major issue, but if the agency has after hours operational requirement (AFP, Customs, Immigration, etc) a forced outage has the possibility of causing major embarrassment for both the agency and the government... Having worked in operational agencies, often scheduled after hours outages are cancelled with virtually no notice due to operational requirements that cannot be predicted.
Response:We agree with your observations on the difficulties of scheduling outages amongst organisations with divergent operational requirements. The majority of our targeted agencies that will use the DCaaS facilities have similar availability requirements and use similar desktop applications. That should enable the co-ordination and scheduling of required maintenance periods.

Question 2 of 5: Are there any other suggestions that you think should be included?:
Where you have stated "all services will be delivered based on formal service level agreements" Will each agency negotiate their own SLA or will it be a case of one size fits all? If each agency is to negotiate their own, the impost on the service provider could be quite onerous, and would become an administrative nightmare. This in turn could stop the use of a shared service in favour of running a large number of individual services, and thus removing any cost saving or service improvements.
Just run the scenario, vendor X offers a service through DCaaS and has 20 agencies sign up, all with different SLA's and contracts. The administrative cost of such an undertaking is great and the vendor tries to pass the costs back to the agencies, who refuse. At the end of each contract the vendor refuses to renew on separate SLA's or contracts. Agencies are then faced with the prospect of using a service that only partially meets their need or to bring the service back in house after getting rid of all of the expertise, licenses and hardware required to run said service. (Sounds a lot like the whole outsourcing fiasco that was forced on to the APS)

Response:The intent would be for all agencies co-hosted in the DCaaS facilities to be covered by a standard set of terms and conditions including core service levels. Where required an individual agency may negotiate an extension to the standard service levels if required.
By consolidating the diversity of agency requirements into a common service approach may be an effective way of managing agency expectation and providing the service provider with more realistic set of service delivery objectives.

Question 5 of 5: What factors do you sees as leading to success or failure in the provision of DCaaS to Government. This may include contract terms, service level agreements, data security and mobility, intellectual property or data ownership.:
Until the WOG DC deatils and papers stating how it will work, answering any of the questions above is at best a general indication, at worst completely wrong.
I am getting the feeling that this paper is a little early, and needs more thought put into exactly what DCaaS is. What services are going to be offered, and at least some high level details of how it will work. Asking people to indicate what they would sign up to irrelevent as without details you may as well ask if people want to sign up to "something".

Response:We opted for this level of detail so as not to dictate the range of solutions considered.
The shape of DCaaS and how it will operate will to a degree be shaped by information shared here on this blog in addition to the other research we are undertaking. The example service offerings are used as catalysts to bring out alternative ideas or approaches that may be viable. Asking if organisations would consider offering a given type of service is a way of measuring likely market interest in providing the services.

Its great to see the government embracing new delivery models and will definitely allow IT services for departments and agencies to be delivered in a more flexible and agile manner. It is already being used in the US and look forward to this being a reality in Australia.

We are conducting a study for Sydney Water into the correlation between Data Centre water and energy use. A particular focus is being placed on those energy end uses that generate heat, what improvement measures are possible and suitable KPIs to measure performance. We have adopted PUE for power and WUE for water. I see no reference to either in your draft Data Centre Spec nor the requirement for measuring infrastructure needed to calculate and report them nor any targets to be achieved. I suggest you review this area – I am happy to assist.
Robert Quinn
Managing Director
National Project Consultants Pty Ltd
0299068611

On
*ICT Infrastructure Policy
*Cost Avoidance Measures

Skills, Technical Support and Quality should also be a benefit. I have recently subscribed to the PM's blog feed to discover it has not been configured correctly - nor apparently tested.

The sender field displays HTML code and the blog itself does not have an appropriate title.

see
http://www.pm.gov.au/rss-feeds/blog

Marghanita

Hello Kayelle,

US strategy:

Adopt Light Technologies and Shared Solutions. We are reducing our data center footprint by 40 percent by 2015 and shifting the agency default approach to IT to a cloud-first policy as part of the 2012 budget process. Consolidating more than 2,000 government data centers will save money, increase security and improve performance.

http://www.whitehouse.gov/blog/2010/11/19/driving-it-reform-update

Comments on this post are now closed. Please let us know if you would like to discuss this post.

Last updated: 29 July 2016