You are here

You are here

Why cloud operations management is the next big thing

public://pictures/davidl.jpg
David Linthicum Chief Cloud Strategy Officer, Deloitte Consulting
 

This is the year to get down to business with cloud computing. It's time to move from the hype-driven world of cloud migration—including application modernization and digital transformation—to the humbler world of long-term cloud-based systems operations. 

Here's where many IT shops stand today: Their organization moved to the cloud in a big way, and they now have a mix of cloud configurations—whether they call it hybrid IT, hybrid cloud, or multicloud—and the looming X-factor is how to manage the resulting complexity. They're ready to set up standard operating procedures for long-term cloud operations. 

It's already clear that rapidly expanding cloud ecosystems are shaping up to be a huge headache for IT in the coming year. Success depends upon having an effective CloudOps strategy. You need an action plan to deal with everything from problem management to monitoring, security, and performance.

Here's how your team can better control cloud management issues at scale, including what needs to be done now to ensure success

CloudOps requires a reset

Here's an uncomfortable truth: Most enterprises that migrated to cloud-based platforms, both public and private, did not focus on operations. Most transition teams believed that operations would be covered by the existing operational teams and that cloud computing systems would simply be another platform to be maintained, secured, and governed. 

The enlightenment usually happens when enterprises are 30% to 40% migrated, both data and applications. Teams discover that approaches and technologies leveraged for traditional operations are unsuitable for use with cloud-based deployments. 

If you're wondering why, the answer lies with these three main factors:

1. The remote nature of cloud-based platforms  

Because cloud systems are out of the direct physical control of ops teams, special care must be taken in how they carry out operations. Typically, the ops team is dependent upon the cloud provider for all information, and the provider schedules and executes most platform-related fixes. 

This lack of control actually contributes to the value of leveraging public cloud. You turn over most of the direct control to the public cloud provider, which typically involves relinquishing the time-consuming control of platforms down to the raw metal. In rare instances, it's possible to maintain that control with some public cloud providers, but not many. 

2. Security is a shared responsibility 

Much like the issue above, responsibility for security operations (SecOps) needs to be partitioned between the enterprise CloudOps teams and the cloud provider. And this can get confusing, quick. 

Most breaches that occurred within public clouds did not lack necessary security technology. Instead, there was a lack of understanding about the roles and responsibilities of each party. 

If you want to review the legalese of this relationship, the particulars are spelled out in the agreements you accept (or accepted) before becoming your cloud provider's customer. Today, it's hard to find a CloudOps team member who has read and understands the legal jargon.  

3. Teams can't keep up with the number of new cloud services moving to production

The "abundance of choice" or the need to select best-of-breed are the prime culprits of increased complexity that, in turn, the CloudOps team finds the most challenging. A report by Aptum explains that abundance of choice typically results in a technological smorgasbord, where decoupled cloud dev and migration teams make their own calls around what technology to use.

Complexity naturally arises when it's time to join and coordinate the multitudes of those apples and oranges, which come in every flavor and variety. 

Key tools and technology

It's time to recognize CloudOps as a new, separate discipline. Although many people had hoped to toss CloudOps in with traditional operations, the special needs of public cloud-based applications and data just won't allow that strategy to succeed. Get comfortable with the idea that cloud requires a new set of concepts, approaches, and, yes, technologies. 

Enterprises that have a set of ops tools and hope to extend them out to their newer public cloud platforms will find that the old tools won't meet the changing requirements. Most ops teams will need to invest in new brands and technologies that make cloud and non-cloud operations specialized and separate, and thus more costly than expected.

The good news for CloudOps has been bad news for traditional enterprise IT Ops. No matter what report or metric you review these days, legacy systems do not get the investment they did just a few years ago. 

A report by the Cloud Industry Forum concluded that UK organizations devote 19% of their IT budgets to cloud infrastructure, just ahead of the 18% spent on on-premises infrastructure, and that by 2022, just 12% of IT budgets will be spent on legacy technology, as cloud usage increases. Keep in mind, legacy systems still comprise the majority of systems that run businesses

Among tool-makers, the lion's share of R&D investment is earmarked for CloudOps tool development. Obviously, this new focus comes at the expense of traditional ops tools, where users already see fewer updates and almost no innovation in on-premises systems operations technology. 

The good news for those already in the cloud, or moving there, is that new categories of technology will be created around the emerging needs of CloudOps teams. This will include tools such as AIOps (artificial intelligence operations), CI/CD (continuous integration/continuous deployment), IAC (infrastructure as code), SecOps (security operations), and MLOps (machine learning operations), just to name a few, all created specifically for the cloud.

The current state of the market makes it possible for those who support CloudOps to access special-purpose tooling, where the capabilities of CloudOps tools have surpassed the capabilities of traditional operations tools. But back on premises, the once unthinkable is becoming commonplace: Systems that run the majority of business processes are often the most neglected.  

Steps to CloudOps success

Your plan to succeed with CloudOps should include establishing approaches, playbooks, tool sets, skills, and ongoing processes to achieve sustainable long-term success. 

Here are four emerging best practices that usually translate into an effective, optimized CloudOps enterprise. 

1. Include traditional systems in the CloudOps strategy and tooling, if possible

Typically, this is not easy. There are paths to leverage the same operations approaches, including common SecOps processes and tooling, that may simplify enterprise IT operations overall. 

However, in many instances, existing enterprise ops teams force common solutions and processes into places where they are not an optimized fit. While it's more complex to keep different tools and skills around to support cloud and non-cloud systems, for the majority who approach CloudOps, there are no other options. 

In this step, you're determining if using a combined toolkit or processes is even a possibility. 

2. Promote data to become a first-class citizen  

Data is often mixed in with general-purpose operations and approaches, including data that's managed on public clouds. This is a big mistake. Consider the data—across platforms, across applications, even across clouds. 

Managing data under a holistic and unified approach carries huge advantages. The DataOps strategy needs to be large and overarching, and then broken down into different approaches and tooling. This includes DataOps technology such as master data management (MDM), backup and recovery, data performance engineering, microservices, containers, serverless, and even data using cloud connected edge-based systems. 

However, all should exist under a unified approach to DataOps that everyone on the ops teams understands. 

Proper management of your data, in the cloud and out, is about 75% of the CloudOps battle. 

3. Make security systemic throughout CloudOps 

While SecOps is part of CloudOps, there are many organizations that try to put SecOps, DataOps, NetOps, GovOps, etc., into their own silos. The trouble with that approach is that security needs to be a part of every aspect of CloudOps. Your workload security is only as effective as the most vulnerable component. 

Even though SecOps should be centrally managed, the security systems leveraged in public clouds should be purpose-built for the application, data, systems, use case, etc. Yes, this means you'll spend much more for cloud security than you might have anticipated, and, typically, the most secure cloud-based systems will also be the most complex. 

4. Place volatility and complexity into an operational domain 

The most successful CloudOps shops understand their underlying cloud-based services and how applications and data storage leverage those services. They also realize that cloud-based services will be in a constant state of change, pretty much forever. If you restrict change, the value of cloud computing goes down significantly. 

Rather than limit change, learn to manage it effectively. The best practice here is to figure out where things change the most, such as database and development technology. Then learn how to leverage abstraction and automation to reduce the number of changes that need to be made to CloudOps processes and approaches. 

For instance, leverage data services for all database access, no matter what technology or platform. This approach allows the actual database to change as often as needed, limiting the impact on applications or other entities that leverage data. This can be done for processes, DevOps tooling, and even operating systems. 

Get a clue, or buy one

This year the alarm will sound around the importance of CloudOps. The fact that most enterprises have no clue as to how to properly implement CloudOps will also come to light. 

A few enterprises will react by slowing down their digital transformation efforts, which will allow ops to catch up. But if your enterprise ends up on that path, the above best practices were applied improperly or ignored altogether. You can count that as a fail. 

If your enterprise does follow best practices, you will still need to remain vigilant to keep change and complexity issues under control. This is your mission, and with the proper planning and management, you will succeed.

Read more articles about: Enterprise ITCloud