EMA estimates that enterprises today waste over 50% of their IT budgets on inefficient application workload placement, configuration and management. As a side effect, they introduce tremendous operational risk in terms of security, regulatory compliance, performance and reliability.
EMA’s 2017 hybrid cloud, software-defined data center (SDDC) and machine learning-related research projects (take a look at our prospectus) aim at helping customers address this waste and risk by slaying the four horsemen of hybrid cloud (listed below). Keep in mind that you may need to slay each horsemen hundreds of times—once for each application you need to deploy and manage—before he remains in his grave. This can turn into a real bloodbath.
The 4 Horsemen of the Hybrid Cloud Apocalypse
1. First you make the mistake – it is easy: Incomplete information when matching application workloads with infrastructure environments is the root of all evil and, naturally, has two sides:
- The infrastructure impact of application workloads is insufficiently understood. This means that customers go by the requirements on the box, combined with Google searches, when selecting their server, storage and network infrastructure for a specific application. What the box does not tell them is how sensitive or forgiving the application is when it comes to temporary network or storage latency, CPU contention or a slowdown in the corporate database cluster due to an end of quarter fire sale.
- The cost, performance and reliability characteristics of the target infrastructure is not sufficiently transparent. Using an Amazon spot instance is incredibly cheap, while reserving infrastructure can increase cost tenfold. Avoiding noisy neighbors will cost even more, but despite the extra cost, is not fully possible.
EMA Cloud Rants – What Does the VMware – Amazon Partnership Mean for You
2. Then you are stuck with the mistake: Application workload lock-in is what makes you pay for your mistakes. Typically, once the application is deployed to a certain environment, it has to stay there for at least three or four years. This is for the following two reasons:
- Contractual lock-in: Cloud providers incentivise customers to sign long term contracts to offset the initial hardware procurement and setup cost. Once this contract is signed, the app typically stays where it is.
- Technology lock-in: Once network, storage, servers, hypervisors, firewall, performance monitoring and security are configured for a specific environment, it is costly and risky to move the app.
EMA Cloud Rants – Workload Mobility in the Hybrid Cloud
3. You may not notice the impact of your mistake: After you have made the wrong deployment choice a few (dozen) times, you will have applications on different bare metal, virtual, private and public cloud platforms. But what you do not have is a single pane of glass to see if a storm is brewing and, if so, what this could mean for your business. Instead, you have multiple management interfaces and monitoring dashboards—vSphere/vRealize, multiple Amazon accounts, SCOM, UCS Director and so on—that help you react to the different types of crises that will occur on a more or less frequent basis.
EMA Cloud Rants – A Single Pane of Glass for Your Hybrid Cloud
4. You also do not know what the mistake means to your business: This is the “bonus horseman” that gets you when you are already on the ropes, putting out fires. In a complex hybrid cloud setup with a myriad of deployment and management choices, you will need something that ties your organization’s business goals to the decisions you make in the data center. For example, if “customer retention” is your company’s main goal for 2017, you need to align your provisioning, management and issue remediation decisions with this goal, even if this will cause issues in other areas. The problem is that you need something or someone to tell you which events are relevant for your customer retention goal and which are not. To me, this type of system is the secret sauce that in 2017 will turn the SDDC into the business-defined hybrid cloud. The SDDC needs a “brain” that controls the NOC, the VMware management guys, the DevOps group, the storage guys, the network people, the database nerds and everyone else in the IT organizations and radically aligns them all with your corporate goals.
Let me conclude by saying that I look forward to helping our customer and vendor community slay these four horsemen by providing insights, methodologies and crazy ideas in 2017. If you are curious, go ahead and take a look at my official Research Calendar for Q1.