What were the key lessons consultants and practitioners learned from developing and deploying hybrid IT infrastructures? What were the common failures, and what were the key takeaways?
Hybrid IT techniques and technologies continue to evolve at a fast pace. The vast majority of organizations don't rely entirely on the cloud, but instead keep most of their apps safely on-premises in their own data centers.
The hybrid approach is practical and prudent because it lets companies control which things they do and don't want to expose to the cloud. It provides the ultimate in choice and flexibility. But this mixture of cloud and "on-prem" applications and services can pose interesting challenges to the IT department, and unexpected issues can seriously disrupt or delay cloud migration projects.
And there’s a cost. One is complexity, including in operations, development, and deployment. Because the number of platforms increases, IT must retrofit management, operational processes, and technology to keep hybrid IT running.
Here are the three most common mistakes, from real-world experiences, that could derail your hybrid IT deployments.
1. The wrong workloads and data
Enterprises moving to hybrid IT do so with workloads in mind that they believe should be in the cloud or remain on premises. The problem with this preconceived notion is that the locations of workloads should not be emotional decisions.
The foundation of a successful hybrid IT deployment is the ability to pick the right workloads to move to the cloud, as well as those that will remain behind within on-premises platforms. This is a logical process where the patterns of workloads and data are understood. Then decisions are made as to which workloads to move, and, finally, a plan is made to move them.
Pick the wrong workloads and data, and the negative effects on the outcome of hybrid IT can include:
Compliance issues
In certain cases, the use of a public cloud is not a good idea. While regulations that limit the use of some public clouds for some data types are rare these days, they still exist. Issues include the cloud's ability to support an advanced encryption service or to transport encrypted data while moving some workloads and data types to the cloud.
Performance issues
Applications that communicate between on-premises and the public cloud usually have a latency issue when many Internet connections are employed, or when the applications themselves are "chatty"—constantly sending and receiving data from on-premises or the public cloud.
Operational limitations
A workload in the cloud may be up and running, but there may be no good way to monitor and manage the applications. An example would be applications that leverage partner applications on other public cloud platforms, where you can see the portions of the workload running on-premises and within the public cloud. But the back-end processes that are not part of your infrastructure have no way of being monitored and managed. This will lead to problems down the road, and moving to hybrid IT only complicates the problem.
2. Performance problems that aren't easy to fix
Let's drill down on the concept of performance as related to hybrid IT and how this can become not only a roadblock, but a system killer as well. Performance is a concept that is directly related to the number of components that a system has and how fast those components run.
For instance, consider an inventory control system that has three major components: the user interface, the processing system, and the database. Assume that the workload leverages a network, a compute platform, and a storage platform.
A simplistic performance model might look something like that shown in Figure 1:
Figure 1. A simplistic performance model for an example app.
Within this sample application, as with most applications, the interface is tightly coupled to the processing system, which is tightly coupled to the data. All of the application components leverage the storage and compute services, no matter if they are hosted on a public cloud or not.
Performance issues typically arise when one or more of the components depicted in Figure 1 slow down. They can slow down for normal operational reasons, such as the network becoming saturated or the database not being tuned properly. Or they can slow down for configuration reasons, such as a tightly coupled user interface and processing systems where one part exists in the public cloud and the other part exists on premises.
If you think that's a bad idea, you're right. Most of the performance problems around the deployment of hybrid IT is not a fault with the hybrid platforms; it's splitting up application components between on-premises and the cloud. These are likely to cause performance and reliability issues because you're leveraging a slower network component: the open Internet.
The trouble comes when enterprises leverage hybrid IT in an attempt to "get wiggy" with the architecture, including placing data on premises and the application in the cloud. Typically, this is to get around the distrust that many enterprises have about putting data on the public cloud. However, the result is poor performance and a failed deployment.
The reality is that if the application components are indeed tightly coupled, such as with our example above, they all need to exist either on premises or completely on the cloud. Of course, you can build applications to account for the performance issues of being run on both on-premises and cloud-based platforms. But unless your application is net-new, you will have to refactor most of the code to optimize the application for a hybrid platform. This means time and money that most enterprises don't have available.
3. Technology evolutions that don't go your way
Finally, the larger issue that hybrid IT must deal with is evolving technology, including some updates that may not add value to your hybrid IT deployments. This does not mean that the evolved technology is bad; it may simply be misapplied.
Here are a few examples:
DevOps tools that don’t work well with hybrid IT
Tools often either support on-premises or the public cloud, but not both. The DevOps process, tool chain, and even the culture should reflect the opportunities and advantages of hybrid IT. Picking the wrong tools means that DevOps won't work well, or at all.
Cloud features that upgrade you to failure
While the public cloud providers add many features each week, those updates may be more than you bargained for. "Upgrades" to new ops tools, security tools, and management tools may negatively affect your security system, or even your ability to operate applications within a hybrid IT platform. You need to pay close attention to what upgrades/updates are coming and the things that they might break.
On-premises hardware and software upgrades that also break stuff
If you want to put fear into the hearts of on-premises operations, just tell them that hardware and software refreshes are coming. Such updates often create problems with compatibility and security or have other issues too numerous to name. The only way to combat this problem is to keep a tight control on configurations, including dealing with upgrades and updates that you may or may not need as part of hybrid IT. I know some operators that have stopped updating on-premises systems altogether, choosing to move to the public cloud instead.
Like it or not, new technology is not always a blessing. As we begin the age of continuous improvement to software, this may mean continuous failure for those charged with hybrid IT ops.
The answer is not to stop upgrades and updates, but to pay attention to what these upgrades and updates will affect—in other words, what they touch. We're so new at hybrid IT that it's difficult to spot problem patterns, but most enterprises are learning through their own failure or the failures of others.
Fail fast to win
While hybrid IT is really the wave of the future, it's also problematic. There is still a whole lot more to learn, and it will be a while before best practices emerge. For now, it's best that you learn all you can from the experiences of others, and figure that you’re going to need to "fail fast" to win at hybrid IT.
Keep learning
Choose the right ESM tool for your needs. Get up to speed with the our Buyer's Guide to Enterprise Service Management Tools
What will the next generation of enterprise service management tools look like? TechBeacon's Guide to Optimizing Enterprise Service Management offers the insights.
Discover more about IT Operations Monitoring with TechBeacon's Guide.
What's the best way to get your robotic process automation project off the ground? Find out how to choose the right tools—and the right project.
Ready to advance up the IT career ladder? TechBeacon's Careers Topic Center provides expert advice you need to prepare for your next move.
