Due to the benefits of improved flexibility and resilience, organisations are increasingly adopting multi-cloud networks. However, migrating to – and managing – a multi-cloud network can be an ongoing complicated process.
Managing a multi-cloud platform is further complicated by the increasing number of suppliers that are available, but this also creates opportunities. For example, multi-cloud network architecture is ideally suitable for organisations that have geographically distributed users, some of whom may not be located near a datacentre.
Organisations can easily be overwhelmed by the process of migrating to the cloud and could be tempted to press ahead without considering the issues that might arise. Considerations need to be made as to which suppliers are most suitable, and also how these platforms can be optimised for performance.
“The more platforms you have, the greater the diversity within the multi-cloud architecture, which in turn drives greater challenges at the management level,” says Alex Dalglish, services director of Comparex UK.
Not all cloud platforms are equal
At their most basic level, all cloud suppliers essentially provide a similar type of service, but each supplier is slightly different, for example Amazon Web Services (AWS) offer auto-scaling. The difficulty of the decision regarding which suppliers to use is compounded by cloud providers often operating without knowledge of each other’s activities.
Thus, changes in services can occur without assessing the impact that this will have upon customers that combine cloud services. To counter this, organisations should remain aware not only of their various cloud suppliers’ current operating procedures, but also how these are likely to change in the future.
“All cloud platforms offer the same types of services, it is just what efficiencies you might be able to glean from each different supplier,” says Dalglish. “Each has their own flavour and levels of attractiveness, based on what operating service you might be running or what purposes you might require for those workloads.”
Acting on impulse
Before jumping into the cloud, organisations need to pause to consider their technological needs and what they aim to achieve.
As part of the planning, the entire network paths should be mapped, identifying where the cloud needs to fit within the architecture.
Alongside this, organisations should identify which applications are most suited for cloud-native implementation and match these to the services that play to their strengths.
Locked into a single supplier
Being locked into a single supplier is the antithesis of multi-cloud architecture, as it eliminates one of the key advantages of multi-cloud networks; namely improved resilience and flexibility. This flexibility comes in the form of being able to align each service with the optimum platform.
However, being limited to a single supplier can inadvertently happen unless organisations take steps to ensure that the data they generate remains in a standard format, rather than dedicated to a specific application or platform.
Read more about multi-cloud
- The debate about what is the right enterprise IT architecture has moved beyond private, public and hybrid clouds to the need to run multiple clouds.
- Public cloud users in the UK are happy with the service they get, but a hybrid approach may offer greater flexibility.
It might be easy to assume that hybrid-cloud architecture is just another form of multi-cloud. But, while they do share some traits, they are in fact different. Unlike multi-cloud, hybrid-cloud is a combination of a public cloud and a private cloud.
“Multi-cloud is not locking yourself into one supplier. Hybrid cloud is running partly on-premise cloud and partly on public cloud,” says Michael Allen, vice-president and EMEA chief technology officer at Dynatrace.
Know what to break and where
Once an application is in the cloud, it can be broken down into a series of microservices in order to gain additional benefits, such as increased resilience and reduced update times. “By breaking the monolith into microservices, you can more efficiently scale up and scale down to leverage the benefits,” says Allen.
However, one of the pitfalls is failing to understand where to break an existing application that you are migrating. “You do not want to be breaking the app into lots of different pieces and then distributing the pieces into different containers, potentially running on different servers, if there is a high affinity between those two different pieces,” he says.
Multi-cloud networks do not make things easier
Despite the many advantages that come from operating in the cloud, multi-cloud networks are not without their own challenges.
Managing a cloud environment is just as involved as managing a corporate network. Adding multiple cloud suppliers makes this even more challenging, as well as increasing the cost and management resources.
“The diversity a customer has in cloud platforms is directly proportional to the amount of supplier management burden on IT,” says Dalglish.
To manage employee time more effectively, companies need to automate low-level tasks to allow their IT teams to focus on the more critical duties.
Failing to automate
Some traditional organisations were struggling to manage conventional application stacks, but the level of complexity that multi-cloud architecture introduces goes beyond what’s possible for humans to analyse.
“You have to have an AI-based system to analyse the billions of permutations for problems and reports to auto-diagnose root-cause,” says Allen.
Organisations need to automate low-level monitoring and maintenance tasks, as well as automating policy across cloud environments and creating exception reports (incident records of errors) to highlight trouble spots.
Once an application is broken down into a series of microservices that are spread across various cloud platforms, this provides greater flexibility and reduced times in updating each microservice. However, it also runs the risk that one microservice could disrupt other microservices due to incompatibilities.
“You now have more moving pieces than ever, you have code changing more frequently than ever, and you have automation moving this stuff around and killing it,” says Allen.
Development teams need to ensure each of the microservices that form their cloud application do not cause incompatibilities and cascading failures.
Multi-cloud networks are complex and resource-consuming, but the advantages they bring – such as increased resilience and gaining multiple supplier-specific benefits – can far outweigh their costs.
However, to take full advantage of their benefits, organisations should take care to avoid potential pitfalls in their migration and management.
“The biggest failing is not being aggressive enough to disrupt yourselves and move faster,” concludes Allen.