If there is one thing that COVID-19 has done for IT, it’s exposure to the woefully underprepared IT strategies. This goes so much further than just enabling remote working. It’s a fundamental building block of enterprise-grade solutions, and the very benefit marketed for Cloud computing. It is of course availability and scalability.
The outbreak of COVID-19 caused a huge demand for Cloud services. The most notable hit was Microsoft Azure. Check this article on the register for more information.
The effects for businesses using Azure was the inability to provision services, unless they could demonstrate a link to fighting COVID-19. This was a serious problem for businesses that utilised auto-scaling virtual machines, which might have turned off when capacity was not required, but failed to start when needed! Other cloud providers were affected by the increased demand but not to this extent. As I write this article Microsoft is still heavily limiting capacity.
What was the mistake made then? Well to understand it we should review the best practice for high availability before the Cloud hype. If you were to deploy an application that was required to always be available, you would provide it from multiple data centers and typically have extra capacity to deal with failures or increased demand. The most advanced would look to use multiple suppliers to ensure no single points of failure. The result if one data center was to suffer an issue the second would provide the capacity to keep the application online.
Sadly the above is now often overlooked with responses like ‘It’s in the Cloud’. The Cloud is not a magical solution that means you no longer need to build the right foundations. Quite the contrary the Cloud is simply just a great marketing gimmick. The infrastructure still resides in a data center and still needs to be designed appropriately. The Clouds benefits are really that it enables services to be deployed and managed much more easily. Whilst providing a wealth of tools and services to empower high availability and scalability, you just need to know how to use it!
So how do you supercharge your infrastructure to the promised land of high availability and limitless scalability? Well simply you look to leverage the number of providers in the market and the number of locations they make available. In an ideal design you lay the foundations of a network layer across the multiple Cloud providers and then deploy your services appropriately. If you design the solution with the correct architecture the prices will often only be slightly more expensive. This is achieved with auto-scaling infrastructure, infrastructure as code and well thought through infrastructure. That’s a lot of infrastructure!
A really simple explanation of the above is well depicted in hosting a website. Imagine if you will, that you have two Linux web servers. Perhaps they sync their files to ensure the content is the same, and they can internally interact with one another via a virtual private network spanning two cloud providers. You then provide access to the website via a service like Cloudflare, providing load balancing and importantly verifying the server is online providing your website. If one Cloud provider has an outage, your users remain unaware as the load balancer ensures traffic is directed at the server still online.
This is a super simple example, but it illustrates the point. If you were going to deploy the application to be highly available, you would have had two servers at a minimum anyways. The only additional cost was the inter-connect between the two servers, which is going to be rather inexpensive.
So this is all great stuff, but what does this have to do with the capacity issues that Azure suffered? Well the biggest hurdle for companies locked into the Azure eco-system, was that deploying anything outside of Azure would likely take to long. However, if they had designed a multi-cloud infrastructure backbone from the start they could have spun up services in minutes, not hours or even days. Moreover some customers found their servers de-provisioned and unable to start, married with huge bandwidth restrictions to export data. Even in its most simple form some basic replication to another provider would have combatted this.
In conclusion there is no real reason why companies can’t exploit the saturated Cloud market space to design and implement highly available solutions for minimal cost. Another global disruption to Cloud services is enviable, but a strategy now will ensure you don’t end up another statistic.
Next2IT can provide access to Azure, AWS, Google Cloud and IBM Cloud. All with a single bill and help you achieve Cloud greatness, with a robust highly available Cloud platform. Get in touch to find out how we can ensure you’re ready for the next global disruption.