Premise. During 2019, enterprises will initiate strategies to bring the cloud experience to high-performance, high-value traditional applications (HVTA). This stands in opposition to the popular notion that all data and applications will migrate to the public cloud. Wikibon believes that enterprises should follow the simplest, easiest, and least risky path to bring cloud benefits to their traditional application operations.
All enterprises are in the cloud, to a greater or lesser degree. They may only be using a simple SaaS capability like giving users cloud-based storage, or they may have moved their entire IT function to the cloud to be more like a digital native company. But the cloud is a ubiquitous and permanent feature of business today and will only become more so as business further pursue digital business opportunities.
However, one broad application domain presents special challenges to an enterprise’s cloud strategy: What to do with high-value traditional applications (HVTA)? And these applications do not represent small challenges. These are the systems of record that capture orders, close the books, and comply with reporting laws to keep executive teams out of jail. They’re the manufacturing plant systems that turn raw materials into consumable items that produce revenue. And they’re the supply chain systems that minimize inventory costs, perishability, and theft. In other words, they’re the systems that run the business: if they go down, the business goes down.
The popular notion among many cloud sages calls for all applications and all data to move to public clouds. The problem is that HVTAs often are based on technologies that aren’t offered in public cloud forms or require capabilities not easily provided by public cloud services. Moreover, HVTAs typically are applications that cannot go offline for extended periods without damaging business results and brands. Generally, Wikibon clients have determined that the costs and risks of a lift-and-shift HVTA migration are too great for the business to assume. But they still want to add cloud benefits to HVTA systems.
What kind of benefits? At the top of the list is the option to begin adding “data-first” application services to HVTA systems in ways that don’t compromise the performance, security, and reliability of HVTAs. Most HVTA have a “known process, unknown technology” history. In other words, they were built to support processes like finance, HR, or supply chain that largely were defined by common conventions mandated by law or other types of strict rules; the challenge was to master the mainframe or client/server technology stack required to operate the HVTA. But today’s differentiating applications are “unknown process, known technology.” They are data-first, like AI-driven applications, which will behave and perform according to patterns in data, but that we know will operate in cloud instances. Examples include recommendation engines, voice recognition systems, and advanced cybersecurity services. Cloud capabilities will make it easier to add HVTA data to the data-first mix and provide data-first services back to HVTAs to improve automation and performance predictability.
Wikibon believes that instead of moving HVTA systems and data to the public cloud, the cloud should come to HVTA systems and data (see Figure 1). To do so, every enterprise will follow their own unique path because every enterprise feature unique combinations of business objectives, HVTA portfolios, and appetites for risk, but Wikibon’s research has shown that successful paths forward typically focus on:
- Achieving a multicloud operating model based on public cloud concepts. Public cloud and cloud-native enterprises have conclusively demonstrated the superiority of the cloud operating model in digital business. The enterprise goal should be to adopt this model for private/public multicloud exploitation capable of running HVTAs.
- Taking the simplest path to modernize digital assets. The sexiest approach to modernizing digital assets (including applications) – migrating to native cloud resources – is not necessarily the best approach for HVTAs. Indeed, traditional applications are boring because boring is better for applications that cannot go down. To ensure continued health of enterprise-specific digital assets, like data sources and applications, most enterprises should evolve their HVTAs by targeting change at the system level, moving to new hardware as vendors imbue system products with cloud capabilities.
- Choosing simpler infrastructure that maximizes cloud affinity. Many HVTAs have operated for decades and they sport a technology pedigree that reflects it. Today’s leading cloud infrastructure platforms converge components and integrate with leading cloud management platforms.
- Transitioning IT talent to accelerate change. IT has to evolve to remain relevant, but enterprises should not throw out the good with the bad. HVTAs often feature performance needs or technological dependencies that must be addressed at device interfaces. Over time, enterprises should take steps to modernize HVTAs, but tomorrow’s objectives shouldn’t compromise today’s operations.
Figure 1. Easiest to Move the Cloud to HVTA Systems
Achieve a Multicloud Operating Model
Technology pundits routinely call for businesses to adopt a cloud operating model without explaining what it is. When it is explained, usually the model looks suspiciously like the set of offerings from one or another public cloud vendor. But that’s not an adequate “to be” description for the end goal of any enterprise’s path to cloud. And without an end goal, sustaining commitment, tracking progress, and achieving the outcomes of such a complex array of changes is impossible.
Because migrating HVTA systems to public clouds often won’t work, every enterprise actually is on a path to multicloud, comprising the use of public clouds where possible and private clouds where public cloud isn’t a viable technical or business option. That means that enterprises must craft and target the business to a multicloud operating model (MOM) for their enterprise. Each enterprise’s MOM will be unique in some ways, in large part due to the HVTA portfolio, but Wikibon believes all will feature 5 classes of capabilities (see Figure 2):
- Rapidly build and test data-first software. HVTA applications were designed to be stable and capable of performing repeated, well-understood processes (e.g., accounting, manufacturing, electronic medical records, credit card transactions) at scale. They still have to do so, but enterprises also want to add the cloud-derived benefits of simpler integration with analytics and AI services, both using HVTA systems as data sources and making advanced application services available to HVTAs. HVTA applications must be brought into the continuous integration/continuous delivery (CI/CD) frameworks to allow for more frequent changes and updates without compromising security, scalability, and performance.
- Automated deployments. Manual processes take longer, cost more to coordinate, and fail more often. For example, Wikibon research finds that manual backup/restore processes can cost a large business $1.4 billion over a 4-year period because small restore failures can cascade to large opportunity costs. Automation can streamline system enhancements and administrative practices, greatly boosting availability and productivity. Automation may cost more today but is much cheaper and better over time because it succeeds more frequently, takes less time, can integrate better security and compliance controls, and reduces administrative overhead.
- API as a consumable service. The essence of digital business is to make assets programmable. The technique is to present things as APIs. Ironically, this includes making HVTAs more “programmable.” While the API approach typically is associated with external-facing applications, selectively using APIs to share data in context between HVTAs and other systems is an essential capability in any digital business transformation.
- Simplified and more common operations. If there’s a place where repeatability and cost management is especially important in a MOM, it’s operations. Public cloud companies don’t drive unit costs down just by buying cheaper componentry, rather they streamline processes and reduce errors, focusing on reliably and assuredly performing value streams tied to delivery, not technology. Technologies, management, and automation technologies that help align an HVTA to a cloud operations experience are extremely valuable to an enterprise. That includes solutions that are based on simpler infrastructure, reduced administration costs, and streamlined updating and enhancing of system resources.
- Data-first control. Data is the heart of digital business – and the multicloud operating model. MOM data must be collected, shared, and acted upon to ensure the MOM can support the digital business’s cost, flexibility, and performance objectives. A data-first control plane must be engineered to ensure sharing of MOM data among the four other MOM capabilities. But data-first requires changes to IT people and processes, too. Most importantly, data-first control requires IT decision making to be empirical, iterative, and opportunistic, even for HVTA systems. This is the basis for both Agile development and DevOps. If IT isn’t data first, then the digital business can’t be.
Figure 2. Relationships of Multicloud Operating Model (MOM) Capabilities
Take the Simplest Path to Modernize Digital Assets
Setting aside users, every system is comprised of data, applications, operating environment, and hardware. Of those four assets, the data and applications typically are the most differentiating and valuable to the business, especially when combined into AI-driven apps that fuel new types of engagement, automation, and operations. However, they’re also the most difficult to migrate to the cloud. To be clear, here we’re talking about migrating, not moving. Data can be moved to the cloud pretty easily, but to sustain the operating environment of HVTAs, for example, the applications have to be moved in tandem with the data and other subsystems, which is the difficult part. That leads to four general strategies for moving HVTAs to a multicloud operating model:
- Use a substitute SaaS. If the business can exit an HVTA and substitute a SaaS option for it, that often is the best choice. But this can be an extremely costly and risky approach that can involve asset write-downs, rewiring of compliance regimes, user defections, and unintended operational risks. However, it can be done. For example, the CRM migration from Siebel to Salesforce was surprisingly smooth for many organizations, but in part that was because Siebel CRM was only modestly adopted in many organizations and thus woefully underdelivering value. When possible, moving to a substitute SaaS usually is the best business option.
- Converting the application. If no SaaS substitute is available or viable, many businesses undertake a major lift-and-shift of an application. For read-oriented applications that don’t embed and engineer processes and data structures deep into subsystems to improve scale and throughput, application conversion to the public clouds may be the best approach. But for the vast majority of HVTAs, to repeat, converting the application is the least attractive business option.
- Virtualize and slide. Over the past 15 years, many IT organizations have been virtualizing their hardware assets, thus limiting the coupling between application and infrastructure. For applications running in virtual machines, like those that operate on VMware, businesses can “slide” their applications to public cloud resources that support identical virtual machine environments. This is tantamount to migrating their control software. For example, both IBM and AWS are among the public cloud providers that have deep partnerships with VMware and can provide near-identical VMware systems in their clouds. However, many HVTAs feature performance, scale, or technology dependence that require direct access to underlying system resources, thus limiting virtualization options.
- Add cloud capabilities to hardware. If no SaaS substitute is available and migrating more differentiating digital assets like application and operating environments isn’t viable, the final option is to undergird applications with hardware that is imbued with cloud-like services. Essentially, this is like keeping the application in place and “migrating the hardware.” If enterprises want to align their HVTAs with their MOMs, this often is the best business choice because it targets change to in the least invasive way – at the hardware level. Wikibon believes this will be the favored choice for many HVTA systems: Essentially, it means setting up an HVTA private cloud and providing appropriate data and management integration with other cloud-based assets through “cloud-competent” interfaces.
Choose Simpler Infrastructure that Maximizes Cloud Affinity
Since most HVTA systems will follow a “migrate the hardware” option to cloud operating models, enterprises should start exploiting system options that are most aligned with their strategic MOM. Wikibon calls these options “true private cloud” (TPC) systems. Essentially, TPC systems provide the cloud experience on-premises or near-premises (i.e., in a local hosting facility). That’s ideal for HVTAs that are too difficult to migrate at the application or operating environment level.
The TPC market is evolving rapidly as new technologies, standards, and vendor business practices emerge. By 2020, TPC systems will provide a true substitute cloud experience on-premises or near-premises. In the meantime, enterprises should prioritize system hardware options that offer:
- Converged infrastructure. The baseline for TPC is converged infrastructure (CI), which combines compute, storage, and networking technologies into a single, engineered package that leverages support hardware, service interfaces, and common enhancement cycles. Different CI packaging offers different classes of application support, including scalability, operating environment compliance, resiliency, workload optimization, efficiency and protection. For HVTA systems, CI solutions also offer a range of virtualization options, as may be required by HVTA performance demands or technology dependencies.
- Cloud operating model affinity. CI by itself improves MOM affinity by simplifying and unifying key tasks in infrastructure operations. But the best CI systems go further by providing direct support for two other MOM capabilities: automated deployments and data-first control. Today’s CI systems are being instrumented to capture system data on component performance and feeding it to management applications for superior monitoring, faster and surer enhancing, and what-if analytics. These are the essential MOM capabilities for HVTA systems: awareness of system behavior, automation of capacity changes, and analytics for planning.
- Integration with an advanced MOM control platform. MOM affinity is good, but public cloud providers are driving the cloud experience. Ultimately, enterprises want their HVTA systems to appropriately plug into a MOM platform that adds rapid build and test capabilities and API externalization across an entire digital asset portfolio, including HVTA and other applications and data sources. This is a nascent space that will experience dramatic invention over the next few years, but the VMware software suite is among the leading options today. In particular, vRealize is emerging as a key cloud management platform and should be the starting point for any enterprise with significant VMware investments.
Transitioning IT Talent to Accelerate Change
IT professionals can be a major barrier on an enterprise’s journey to a MOM. Why? Because people are hired, trained, paid, and managed to ensure application delivery, especially for HVTA systems. If the professionals responsible for achieving HVTA performance thresholds feel that a technology will compromise scalability, reliability, security or some other HVTA attribute, they will mobilize to reject it. This happens so frequently that sometimes business leaders conclude that IT doesn’t share the business’s overall digital agenda. While that may be true in targeted cases, Wikibon’s research finds that IT professionals generally act in concert with digital business strategies if their goals encourage it and the technologies they manage facilitate it.
