Antebellum have recently been involved in one of the largest and arguably most important deployments to cloud infrastructure for the Victorian Government. But how did it happen, what challenges were faced and why is ‘cloud’ such a misunderstood and misaligned term within Government?Firstly the term ‘cloud’ is very misunderstood within the major project space and particularly within the Government. We have had CIO’s explain how data centres outside of the Government data centres (CenITex) are cloud, IT departments explain how they can offer cloud computing services and how using platforms such as Salesforce or Office365 is the cloud deployments they understand. The reality is that unless you are consuming software as a service (SAAS) like Saleforce, Xero, Office 365 or the hundreds of other software applications that are delivered across a web browser, as a service then in major projects we are talking about infrastructure as a service (IAAS) or platform as a service (PAAS) from the large cloud providers (Azure, Amazon).

While it is true that most of our software tools such as test platforms (TestRail), communications (Google Apps, Slack) or the multi-purpose JIRA, online planning tools and the like are all SAAS in our major projects. This is because they are cheap to run, easy to deploy (specifically under a Department network environment where getting client software on user desktops in akin to pulling teeth). They make running teams more effective, more efficient and can be utilised on personal devices, vendor devices and Departmental computers simply and easily.

For major project deployments we are talking about IAAS or PAAS. For the latest major project Antebellum has been involved with (the Births, Deaths and Marriages Core System Replacement) we knew from previous major projects that we needed to push for IAAS in a major cloud provider (in this case AWS) but how we would start and what obstacles would we need to overcome?

Initially we established at least the option that we may deploy into IAAS cloud and so from the initial procurement we wrote it in as a requirement that any future system would need to be”cloud aware” meaning it would need to be able to be delivered via web from IAAS infrastructure but more importantly all modules and middle-ware would need to be cloud aware and the solution as a whole would need to be able to scale both vertically and horizontally across regions inline with the way IAAS deployments work.

We then had to prove beyond a doubt that putting such a critical core system into the cloud was a viable option in terms of cost, reliability, scalability and security. With some very smart people we developed a very thorough Options Analysis looking at 3 differing options – internal hosting with CenITex, external hosting with another commercial provider and then utilising IAAS through Azure or AWS platforms. The stark reality was that for new deployments, providing the application was cloud aware there is not a single part of the analysis that is superior to IAAS on the main cloud providers and particularly not hosting on the Government shared service provider. In fact the delays, costs, scalability and time to deploy is so far advanced in IAAS that there is no factor or reason to host on Government data centres. Couple this with the vastly reduced costs even before you take into account billing for use across multiple environments (development, test, production etc) the only viable option is IAAS. Plus the available products and services available through the IAAS platforms are vastly superior, cheaper and a greater range available and most billed as a usable service rather than legacy licencing models.

But still there is major confusion particularly in Government circles as to what cloud is. Our projects technical lead described it simply “if you can run up 100 virtual machines in less than 10 minutes then that is the cloud we are referring to”. While a simple analogy, to run up 100 (or even 10) virtual machines through CenITex would take you at least 6 months and probably cost you in the order of 50 times what AWS charges and there is no elasticity in design. For our major data migration we were able to dial in huge computing power for only days at a time to bring the data migration from 4 days to 4 hours at a negligible cost.

So while the benefits across all areas are clear, what are the challenges? Other than culture and politics, the challenges are architecture and support.

Architecture

In this fast moving world, finding good enterprise or solution architects is difficult but they are available and there is many firms moving quickly to provide services in this space. But more importantly there is very minimal understanding of IAAS within the Departments so when you need to get approval by technical/design authorities and they have very little understanding of what you are deploying, it becomes a major roadblock to approvals and delivery. There is also challenges with bringing traffic back inside the Departmental network that previously would have been contained within the network, and this poses architectural and security challenges for the design (think bringing specialised print traffic back to Departmental printers or trying to integrate with an internal Departmental system from the cloud solution). Changes to Departmental security models and settings (or updating the SOE) are always complicated, timely and no doubt involve copious forms and submissions.

Support Models

The support of IT systems has normally fallen within the purview of Departmental business areas, service desks or the like but since you now have complete applications out on the cloud the entire standard support model changes. Even if you wanted to use internal service desks they generally do not have the skills to diagnose and triage IAAS deployments. While there is no single model, ultimately we implemented a hybrid model using a cloud specialist as the Tier 2 support and internal help desk for Tier 1 support with Tier 3 being the specialist providers (application, network, desktop).

While cloud computing in all of its guises brings some large benefits to major projects and allows for faster, cheaper and a large range of options there are still challenges within certain environments, but it is happening and will continue to move to IAAS and PAAS cloud as legacy models and data centres just cannot complete at any level.