How we put Robots on Cloud Platform Services
Unlocking the cloud’s full potential of scaling with platform services, on a software vendor’s solution that normally wants to run on virtual machines – is that possible? For our Robotic Process Automation product, the answer turned out to be a yes. Continue reading if you want to find out how we managed to do that.
When we started building a shared Robotic Process Automation architecture in the cloud, we wanted to create a web service that is reachable from anywhere, leveraging platform services to minimize the classic operational overhead needed to maintain a park of virtual machines. Achieving this would allow us to leverage
- the superior uptime guarantees
- the much more granular sizing options available for web applications and database services
- the initial and ongoing hardening of these services,
provided by the cloud platform while cutting down costs for building as well as operating this internet-facing web service, which would be used by customers like Entris Banking Ltd.
Based on the vendor’s initial recommendations, we would have built an old-school IT landscape consisting of several clusters of virtual machines, able to power several thousand concurrent robotic processes.
Since we were only launching this service, we were not quite at that level of adoption in the market. Between the architect, the vendor, our managed services team, and of course the product owner, we figured out a solution that met our requirements. The result was an architecture where nearly all components were replaced with platform services:
- The cluster of virtual machines powering the web application component was replaced with an App Service.
- The cluster of virtual machines powering the database component was replaced with an Azure SQL Service.
- Only a single component running a packaged .msi installer remained to be implemented on virtual machines.
So, even if the recommended architecture for a web & database application lists virtual machines, it may be possible to replace these for example with an App Service and Azure SQL DB. It is a great opportunity to work together with your internal application team, your architects, your managed services team as well as your software vendor. You may just be the first party to introduce this idea, and in the process, break new grounds for everyone involved.
Lesson learnt: Challenge your established ways of thinking about IT systems before starting their implementation in the cloud.
Once you succeed in replacing virtual machines with platform services, you can build a landscape that…
- can start with a small footprint.
- scales up in small increments.
- comes with a superior SLA of 99.95% per component.
A further set of advantages are that this platform-as-a-service based architecture…
- is charged at a fraction of the run rate of an Availability Set based environment with multiple virtual machines per component.
- needs less effort from your managed services team in the years to come.
- allows everyone involved to learn.
Please do not underestimate the last point: the IT landscape is changing dramatically. The global public cloud providers introduced a new way of doing IT, which has become more and more relevant and may well one day be… dominant. If you are looking for a new job one day or want to attract new talents, having used the cloud helps. Especially if you leveraged it in creative ways.
And if you are struggling with the challenge of learning the new approach to doing things that the cloud introduces, maybe Josh Kaufmann’s message can help you, because in reality, it takes far less than the often quoted 10.000 hours to pick up a new skill:
No matter where you currently are on your learning curve, we would love to help you. Be it an ideation session on the whiteboard with our cloud architects, a peer-review of your solution proposal or managing your assets in the cloud: we will support you on every step of your journey to the cloud, just reach out to us.
And for your classic workloads that will remain on virtual machines – cloud or not – stay tuned for ways to save a significant chunk of their run rate and licensing costs in Azure, which we will cover in a future blog in this series.