With hybrid cloud no longer merely a stop along the way to full-fledged public cloud, tackling management complexity is essential.
Cost is among the considerations IT leaders take into account when assessing whether to deploy workloads on the public cloud or on-premise. Tony Lock, distinguished analyst at Freeform Dynamics, confirms “complete cost predictability” with public cloud can be hard to achieve, but cloud allows organisations to achieve things otherwise too difficult or complex to try or even think about.
The trouble with mix-and-match hybrid cloud is defining requirements in relation to specific needs. Users might require a mix of on-site, private cloud, local cloud services sitting in between, or standalone silos serving individual applications, with many adding in public and multicloud too.
This can lead to a highly heterogeneous IT environment. “It’s all about the art of the possible,” says Lock. “You can mix it up with infrastructure cloud providers and past cloud providers, and with SaaS [software-as-a-service] providers as well increasingly, to extract data they’ve got in an on-site application, with information in SaaS, and maybe analyse that with ML [machine learning] algorithms from one or more public cloud providers.”
Major services firms – from HPE, IBM and Dell Services to Microsoft and Google – all “do” hybrid cloud, with public cloud players realising long ago that not everyone was going to leap on “US Cloud X” since many need local presence, data sovereignty or similar, Lock explains.
For instance, Microsoft packages elements of its cloud service to run on a server in a firm’s own datacentre, although these solutions typically act as part of that bigger public cloud, with the customer having full control, including for update authorisation.
This highly heterogeneous IT environment can cause management headaches. What’s crucial, says Lock, is the questions you ask of different suppliers. “Are you going to get the sort of support you need to make this work? Do you have all the knowledge and skills you need to run it – if not, have you got a specialist in mind that knows them well enough and knows you well enough to make that work for you?
“Can current or envisaged infrastructure support a desired solution and service level agreements (SLAs)? Have you exposed all the redundancies across the organisation? What’s running that shouldn’t be? How is your tech evolving over time? What might be done bit by bit as sub-components?”
Organisations may not be in a position to put skilled IT people on two-month training programmes to work out the art of the possible within a specific option, yet you need to understand how to achieve what’s needed, safely, cost-effectively, and with some hope of an optimal result.
Once you’ve answered the questions, you should be in a much better position to pull it off – or choose a different cloud mix. Avoid allowing it to come from the board as a company directive that sounds good, from their perspective, as a goal for 2025 or the like.
IT leaders should also ensure they avoid the temptation to provide department heads with new applications they request. Rather than responding to a request for something like a new accounting system, Lock says IT leaders should ask departments proactively what might prove useful and what they’re interested in.
They should then ensure visibility is sufficient to enable manipulation of the environment to meet changing requirements. “Managing cloud is about process, process, process. Then you can run hybrid cloud in any of the myriad versions or recipes,” says Lock.
Consistent data management across on-premise, hybrid and multicloud environments is another factor to consider. Grant Caley, chief technologist for UK and Ireland at NetApp, recommends IT leaders have a plan on how to store, manage, optimise and secure data in the chosen environment. “What I see customers struggling with is working out how to standardise and operate across that,” he says.
Consistent data protection, data security and provisioning drives cost optimisation, with standardising the layers – building horizontally on top of the a hybrid multicloud environment – crucial for manageability.
“We can’t just have data or security or disaster recovery managed differently on-premise to the cloud. You need consistency operationally to focus on leveraging the services on which these different environments bring to bear”
Grant Caley, NetApp
“We can’t just have data or security or disaster recovery managed differently on-premise to the cloud. You need consistency operationally to focus on leveraging the services which these different environments bring to bear,” he says.
Once some of those challenges are solved, start to think about how to pull in helpful cloud or on-premise services, devising an answer that fits and helps ensure that all the different requirements, processes and procedures for the environment relate to each other.
At NetApp, SaaS such as Microsoft 365 or Dynamics might be chosen if it reduces the complexity of management, he says.
“But you’ve got a lot of middle ground – maybe having written your own applications, or have been going into containerisation and paths or platforms that can be on-prem or cloud,” says Caley. “Then it becomes an argument about all those other factors. Have you got skills? What’s the cost profile? What do you really need? Do you need cloud AI [artificial intelligence]?”
However, he adds that “true” hybrid multicloud, including the building of multicloud apps spanning different public clouds, remains fairly rare.
What about multicloud?
“The risk of building across multiple clouds with different service levels and connectivity in between to make sure they all work together may still be too big a challenge,” says Caley. “More likely, it’ll be about containerised applications and moving whole pods of applications, including all their dependencies, to a new environment.”
This can all be largely dependent on what has been inherited, either through organic organisational growth or by blending with other IT environments through acquisition or merger.
Arun Chandrasekaran, distinguished vice-president analyst at Gartner, suggests hybrid cloud might be best understood more as an underlying philosophy than a specific market with specific technologies.
It might also be speculated that this reflects that comparing hybrid cloud environments is tricky – mixing apples with oranges, bananas and a few lemons, or mangoes too.
He agrees that hybrid cloud requires a “consistent management and governance experience” across the estate, with consistent security, including non-siloed security policies as a “first and foremost pillar” of hybrid cloud management.
“We’re starting to see a rise in more application layer attacks, targeting the software layer or even APIs [application programming interfaces]. Attack vectors are kind of moving up the stack,” warns Chandrasekaran. “You don’t want this quagmire of haphazard tools that you’re using across different environments – and you now have to hire specialists to manage these across multiple environments.”
Secondly, organisations should look at stability and monitoring the environment – make sure you can understand how workloads are functioning and what’s causing any downtime. Observability and monitoring are crucial, he says.
Next comes networking; after all, hybrid cloud is all about connecting different clouds. You need solid layer two and three networking, more application networking, and services must communicate with each other.
Chandrasekaran says the fourth pillar to manage should be cost optimisation, which has become increasingly crucial given the troubled economic climate. Beyond that, attention to data storage, backup and disaster recovery policy, and provisioning, with automation to remove friction – enabling the provision of infrastructure in an automated way, including across developer pipelines.
Public cloud providers have made efforts to support other clouds, such as Google BigQuery or Microsoft’s third-party management plane. They’ve tried services that enable hybrid cloud – for example, the Amazon mini-appliance Outpost for customer datacentres, or Microsoft’s Azure Stack, or Google PaaS.
“My opinion, bluntly, is that none of these efforts [things like Outpost or Azure Stack] have been wildly successful,” Chandrasekaran says. “[Although] customers are constantly telling you they want cloud, these appliances themselves haven’t gotten a lot of adoption.”
James Sturrock, director of systems engineering at Nutanix, agrees on the need for monitoring and management tools, and interrogating all the aspects of what an organisation wants and why for maximum agility and flexibility.
“If you want to move an application from one place to another, can you just plunk it there or do you have to look at it and transform it, into SaaS or something else? What are the dependencies?”
With that in mind, Nutanix offers a management plane aimed at avoiding “nightmares of having to stitch it all together”, he says. These are horizontal management layers that deliver visibility, help deliver understanding and moveability, and can cover just a single node, or edge locations, with flexible pricing available, says Sturrock.
Similarly, Omar Khan, general manager of Microsoft Azure product marketing at Microsoft, says the supplier is building on work started with Azure Arc for centralised management and control of heterogeneous environments via a “single pane of glass”.
“Azure Arc offers cloud services like Azure Data Services, AI/ML and Microsoft Security a consistent set of tools and services from cloud to edge. Azure Arc now has over 28,000 customers, doubled year over year, including organisations in every industry, like ABB, Greggs and the World Bank,” says Khan.
Assuming IT leaders require a number of providers for public and private cloud hosting, then, as Freeform Dynamics’ Lock points out, rather than selecting from leading brands, they may be better served by broadening their view to think about exactly which service providers, consultancies, mainframe firms or beyond will suit the specifics. This includes comparing different internal clouds, or on-site versus off-site.
Whatever mix of internal and externally hosted IT infrastructure an organisation ends up with, observability and monitoring across this diverse IT landscape is a crucial component of a hybrid cloud or multicloud IT strategy.
Read more about hybrid and multicloud management
Cloud technology can pose billing, management and compliance issues. Here are five reasons why repatriating cloud workloads back on-premise might be a better option.
We look at why the cloud is not always the best choice, including for reasons of cost, application suitability, management, data protection and the needs of the business.
