In the current economic environment, uncertainty is causing organisations to relook at where they can reduce cost. Recently, we’ve seen that the topic of Cloud repatriation has resurfaced for this reason which is also usually accompanied by an example of a failed attempt to move infrastructure to the Cloud.
It’s rarely a balanced argument, so we wanted to share some of our experiences to provide a broader perspective.
We don’t think that Cloud repatriation as a concept is up for debate in terms of whether it happens or not. We’ve seen several customers move workloads from the Cloud back to on-premise infrastructure because it’s not worked for them. The bigger question is why it’s not worked.
In our experience, there are several reasons, aside from a general change in direction, why organisations revert to on-premise infrastructure or initial deployments in the Cloud fail:
Cost
Probably the most common reason organisations move workloads back to on-premise infrastructure is cost. The business case, or perceived cost reduction for a move to the Cloud, don’t materialise for those that adopt a lift-and-shift approach.
The Cloud providers seem to get the blame for this like there’s been some sort of false advertising telling people that it’s going to be cheaper.
From our perspective, the cost of Cloud services isn’t the problem here – instead it comes down to the planning or simply a gap in understanding of what’s required to deliver an overall solution are. Aspects like network connectivity and cyber-security measures can easily get forgotten when it comes to design and cost planning. Management systems, training, and the general cost of change can also play a big part in whether the project is a success or failure.
For virtual machine workloads that are being migrated like-for-like to the Cloud, the total cost of ownership can be upward of 3x the price for using modern on-premise infrastructure models from HCI vendors like @Nutanix when you model that cost over time. Cost-saving often doesn’t come without re-platforming to use cloud-native services, so the cost of change needs to be considered.
Data Gravity
Data Gravity is a critical aspect when it comes to migrating workloads to the Cloud. Whether your applications are onsite or in the Cloud, keeping your data close to the application to remove any network latency issues is crucial. You can’t move services to the Cloud and then have users experience poor or worsened performance.
With data volumes growing, making large data sets accessible in the Cloud can be time-consuming and expensive, either remotely or through migration. In some cases, the time it takes to move that data can make the business case unworkable, and it also needs to be considered when you think about an exit strategy.
Governance / Security Changes
In one of our previous blogs, we discussed that the responsibility for securing your data remains yours if you move it to the Cloud. The providers themselves take zero responsibility for that, which is surprisingly still a misconception some people have.
We’ve seen a few occasions where data has been moved to the Cloud, only for the organisation to realise that they can’t meet the security requirements they need to meet or that they haven’t accounted for the cost of securing their infrastructure in the Cloud.
Regulatory requirements continue to change and evolve, as we’ve seen with GDPR and SchremsII, so having the proper visibility and controls of your data in the Cloud is critical to success. Existing/traditional toolsets will likely need to be supplemented or refreshed.
Up until this point, we’ve spoken about reasons why a move to the Cloud can go wrong. One of the reasons for writing this is because we don’t believe that the Cloud is bad – people aren’t repatriating services because the Cloud has no value. People are moving services back on-premise because they’ve not understood it in the first place, their planning has been poor or done without all the facts, or they’ve put the workloads in the wrong place.
So what’s the upside? What can a move to the Cloud bring you and what do you need to think about to get it right?
Right Workload, Right Place
In our opinion, there are some no brainers when it comes to using the Cloud. One of the biggest is the use of object storage as a platform for our unstructured data.
Often, organisations will store at least 5x the amount of data they need due to the protection schema and the replication and duplication of data. This overhead is reduced when using an object storage platform.
Similarly, nobody ever looks at the profile of their data – if you look at how much of it is active, in our experience, that will be around 20% of the overall data, yet all 100% of that data lives on expensive spinning disk, or worse still SSD.
There’s a tremendous amount of efficiency that we can gain by using object storage services like AWS S3 for our file data.
By combining large scale object storage which is highly available and durable (and cheap) for the 80% of the workload that isn’t active, with a higher-performing storage system for the 20% hot data, we can deliver the best of both worlds and reduce costs by anywhere up to 80%.
For many customers, the ability to rationalise applications and move them to SaaS platforms like O365 or Workday or Salesforce, the virtual infrastructure needed shrinks dramatically too.
Agility and Scale
It’s not always about cost – sometimes, it’s about pure scale and flexibility. For certain services or platforms, Cloud makes total sense – if you’re building an online digital platform to reach millions of consumers across the world, why would you try to build that yourself? If you have large, unpredictable workloads that grow and shrink at will – why would you not consider the Cloud to deliver that service?
Agility is a critical component for some of these services. The ability to scale up and down on-demand, with global network links to reach your users wherever they may be, would benefit from a Cloud model. It doesn’t mean it’ll be cheaper to buy, but it will be much easier to manage.
The Cloud – good, bad or otherwise?
From our perspective, there is no right or wrong answer here – it comes down to putting the right data in the right place for the right use case. Cloud services can provide many benefits when planned and used correctly, becoming even more relevant in today’s enterprise, as users connect to corporate data from outside the office’s confines.
We’ve worked with several organisations to help them plot their way to the next iteration of their infrastructure, whether that be on-premise and in the Cloud, but the thing that’s common across each of them is that there’s no one answer.
More often than not, hybrid and multi-cloud is the direction of travel, utilising technologies like VMware cloud and Nutanix’s Clusters service as the underlying infrastructure becomes more transparent.
The other area coming more to the fore is leveraging other Cloud services once the data is resident there. In platforms like AWS, you can present data to services such as Redshift for analytics or Fargate for stats analysis.
Cloud has its place when used correctly, but there are still many misconceptions about things like the shared security model and data sovereignty. Getting to grips with the cost model is another issue that people struggle with. There’s still education needed regarding the art of the possible and what’s involved in building out an end-to-end solution. Hybrid and on-premise infrastructure is developing, too, with more and more capability coming into on-premise infrastructure to bridge the gap between the two.
You need to ask yourself why you are running the programme and what you hope to achieve from it, both technically and commercially, before going straight into migrations to the Cloud.
If you’ve read this and you’re interested in understanding how we can help you make the right decision when it comes to your infrastructure, you can get in touch with us:
If you’re not ready for that, why not check out one of our case studies and gain an understanding of some of the customers we’ve helped on their transformation:
