Micro Focus is now part of OpenText. Learn more >

You are here

You are here

How to ease your multi-cloud management headaches

David Linthicum Chief Cloud Strategy Officer, Deloitte Consulting

Multi-cloud and hybrid cloud infrastructures are now the norm. And while they have introduced more complexity, they also create more opportunities for you to deliver value to the enterprise. Unfortunately, few enterprise IT organizations know how to manage multi-cloud and hybrid cloud over time, or how to assemble the right tools to enable proper management.

Most IT operations management professionals lack the knowledge and experience necessary to do proper planning, choose the right available enabling technology, apply best practices, and so on. So, what's a cloud operations manager to do?

Not to fear. Here are the ins and outs of Infrastructure-as-a-Service (IaaS) multi-cloud management, including how to plan, how to pick technology, how CloudOps works in a multi-cloud world, and what operations and tooling you'll need to be successful the first time.

Welcome to a multi-cloud world

The writing is on the wall. According to the "State of the Cloud" report from RightScale, "A multi-cloud strategy remains the preference among enterprises even as the percentage of enterprises who use multiple clouds dropped slightly," to 81%, from 85% in 2017. Those planning a hybrid cloud strategy fell to 51% (from 58% in 2017). However, the study said, "There was a slight increase in the number of enterprises that are using multiple public clouds or multiple private clouds."

The reality is that most enterprises will leverage more than one public cloud, and, in some cases, enterprises will leverage a single private cloud and more than one pubic cloud. Thus, the complexity of the pattern will be something you'll have to deal with. 

You'll need the ability to abstract yourself from this complexity and leverage concepts such as automation to make the management and operations of a multi-cloud deployment possible. The rise of complex cloud architectures has caused enterprises to look for solutions that simplify monitoring and management for multiple clouds.

According to Forrester’s cloud predictions, the global public-cloud market will jump from $146 billion in 2017 to $178 billion in 2018. Beyond this year, Forrester predicts it will continue to grow at a 22% compound annual growth rate. 

This is huge, considering that back in 2003 the public-cloud computing market, removing SaaS, was just under $50 million. Looking at the growth of computing in general, this is an explosion in the use of public cloud–based resources. 

Also noteworthy, according to the Forrester report, is that public-cloud platforms represent the fastest-growing segment of the public-cloud computing market, and they will generate $44 billion in 2018.

What’s more, nearly 80% of companies studied plan to have more than 10% of their workloads in public-cloud platforms in three years, according to McKinsey's 2017 research on global cloud cybersecurity.

So most enterprises will have more than one public IaaS and more than one SaaS cloud in operation. Moreover, more and more will have three or more public clouds in operation, including AWS, Microsoft, and Google. 

For example, a company might leverage 400 AWS cloud services as part of its cloud services catalog. It might also leverage 200 services from Microsoft and 100 from Google, for a total of 700 services under management. 

To complicate matters, many of the services will have very different attributes, in terms of costs, means of provisioning, and patterns of the service. However, others, such as storage and compute, may seem very similar in function, and even confusing in terms of which is used for what and where. 

You might have 700 cloud services under management within your multi-cloud solution, but no common guidance as to where the solution operates, and for what purpose.

The new tools are good, but in different ways

Of course, the gut reaction to any complexity issue is to toss tools at the problem, and the complexity around multi-cloud is no exception. There is good news and bad news here.  

The good news is that the tools are around, they do work, and you just need to understand what they are and where they are running. It's not a good idea for you to move to multi-cloud without a good tool platform, including plans, requirements, testing, and ongoing operations and technology updates. 

These tools include cloud services brokers (CSBs), cloud management platforms (CMPs), and other tools that provide a layer of abstraction between the complexity of cloud services from two or more public-cloud providers. There are also tools that manage, monitor, and utilize those services. 

The tools themselves can add complexity, but if you’re working with the right tools for your needs, they will enable you to be successful with multi-clouds. 

The bad news is that this is a relatively new space, the tools are new, and most enterprises have very little experience using these tools. Moreover, the confusion around multi-cloud management is often led by misinformation from technology providers and an oversimplification of what the problem is. 

The key things you need to understand include:

  • All tools are not created around the same pattern, even if they have the same names and categories. For example, CSBs and CMPs are different in the way they carry out management and operations, as well as in how they provide key brokering services to developers and the application workloads themselves.
  • Security needs to be a consideration, since management tools themselves can be the point of a breach. You need to plan through your security requirements from A to Z before you pick and implement your multi-cloud management tool or tools. This also raises costs. 
  • Monitoring needs to be combined with management. You can't do one without the other. 
  • These tools may introduce new problems that won't exist within a single public-cloud solution, such as performance, ops costs, and the ability to deal with composite services made up of two or more public-cloud services. An example would be if you combine the analytics engine on Google with a database that exists on AWS. Since the two cloud providers operate without knowledge of each other, they often change their services without regard for customers that co-mingle their services with other heterogeneous services from other cloud providers.

Features to consider

Now that you understand what the challenges are, let's look at the processes you'll need to select and implement the right multi-cloud management and monitoring tools. First you need to establish the criteria, or what’s important to and required by your enterprise in a multi-cloud management and monitoring tool. 

Consider the following features:  

  • Service provisioning: The ability to launch a service and allocate the cloud resources that need to support the service. This tool can translate a provisioning request from the console, or an API, and translate that request into the cloud-native API of the target cloud within the multi-cloud architecture. 
  • Service monitoring: The ability to monitor cloud services during runtime to determine how they are behaving against your policy. This includes reporting on what occurs to determine the future use of that service, including reliability, cost metrics, etc.
  • Service performance: The ability to monitor service performance and then log resulting data. Poor-performing services should be blacklisted until the issues have been resolved.
  • Service governance/policy: The ability to define and leverage policies around the use and execution of the service or services.
  • Service orchestration: The ability to orchestrate the cloud service to meet the needs of a core business application or business process.
  • Monitoring analytics: A detailed analysis of how the clouds under management are behaving, key metrics, and predictions around future behaviors such as performance or reliability. 
  • Integration with security: The ability to operate seamlessly with the existing cloud security infrastructure. In most cases, this will be identity access management, but it could also deal with compliance systems as well. 

Keeping these features in mind, there might be other criteria you need to consider for your specific situation. Don't be afraid to augment the list.

The correct process: How to succeed the first time 

The basic process looks like this:

  1. Define your cloud architecture, including the multi-cloud services leveraged.
  2. Create a catalog of multi-cloud services. Make sure to define the right metadata so they can be found, discovered, and used. 
  3. Define performance expectations, security, operations, and other aspects that should be understood before selecting a tool or tools. 
  4. Select an array of tools that seem to fit the needs defined above.
  5. Select candidate tools that meet the basic outlined requirements. 
  6. Test tools using live multi-cloud testing environments, include simulating workloads, composite services, security ops, etc.
  7. Select tools or tools based upon outcome.
  8. Do acceptance testing.
  9. Establish a process for re-evaluating tools on an incremental basis.

Multi-cloud ops: What’s new and what's not 

When considering CloudOps, there's more good news and bad news. 

The bad news is that the new tools will take some learning before they become part of enterprise IT operations. Successful CloudOps requires that IT operations understand multi-cloud management and monitoring tools, as well as the native clouds they manage. This additional complexity leads to frustration, at least at first, especially considering that things already seemed complex at the outset. 

The good news is that there's much more good news than bad. If you take the time to properly define and design your multi-cloud solution, then select the right management and monitoring tools to fit your requirements, your efficiency will go up tenfold. Moreover, the additional agility of your solution will quickly add value to the business by a factor of 1.5 or 2.0. This will pay for any risk or costs that you encounter. 

Want to know more? Read my new white paper, Multicloud Monitoring: How to Ensure Success the First Time

Keep learning

Read more articles about: Enterprise ITCloud