IT operations is evolving in surprising ways. The obvious changes are the ones brought on by virtualization and the cloud. Five less obvious trends, however, are changing the dynamics of operations (or ops), each of which is defined and explained in this article. These trends include configurable ops, lean techniques, accelerated cadence, hybrid architectures and the consumerization of IT. No doubt, a competitive advantage is gained by those IT shops that respond most effectively to these changes and trends.
In addition to the lesser-known trends, this article suggests six responses that, if mastered, optimize IT operations: synchronizing people with processes, systematizing demand management, automating repetitive tasks, integrating people with system processes, rallying around an ops calendar and orchestrating service management. Together these steps will make ops processes automated, transparent, configurable and connected. In a word, ops becomes orchestrated, and orchestration creates dramatic improvements in cycle time, compliance, agility and also accountability.
Five IT Operations Trends
IT operations is rapidly evolving. Virtualization, the cloud, revenue generating apps, web services consumed and delivered—all make today’s ops vastly different than that of even a few years ago. This altered environment has led to new operational realities. Five trends have emerged as vitally important for Ops teams to understand and master.
1) Configurable Ops
Whether hosted on-premises in a private or public cloud, there is an increasing push to make run-time environments deeply configurable. Configurable environments reduce the need for deployments by allowing for meaningful updates to the end-user experience without rolling new code into production.
The need for configurable environments requires that development have significantly greater involvement with ops—one driver of the DevOps phenomena. Whether titled DevOps or not, dev engineers are working closer with ops engineers to design configurability into apps and the platforms on which they run.
2) Lean Techniques
In recent years lean production techniques jumped beyond car manufacturing to other sectors of manufacturing and to service industries like hospitals and restaurants; now they are jumping to IT. The lean methodology emphasizes the removal of waste, where waste is defined as non-value-added activity. Two important manifestations of lean techniques include “just-in-time” provisioning and the use of “pull versus push” production control. In a pull-based operations environment, new releases are put into production quickly on request. This capability only works in practice when release and other regularly occurring operations are honed down to their essential activities and can be executed rapidly and reliably. Undertaking lean operations requires that a suitable request system be used to support pull notices and visual indicators of current work status. For instance, a real-time operations calendar may serve as the visual indicator.
3) Accelerated Cadence
Operations must keep pace with the accelerated cadence of a rapidly evolving online world, especially now that the apps being served out of ops often serve as important revenue for the business. The pressure to constantly enhance critical apps is felt acutely by dev, but also by ops. Because regardless of whether dev formally undertakes the agile methodology, they will surely be sending more-frequent releases to ops and will expect ops to crisply and confidently deploy them into production.
This leads to weekly, daily or even hourly deployment cadences, as opposed to the more stately monthly or quarterly cadences of just a few years ago. There is, however, an unavoidable need to deploy much more often than before. To state the obvious, to deploy much more often is to deploy much faster, leading to much tighter release windows. Thus a cadence of eight releases per day leads to a three-hour release window, assuming an even flow of deployments over the entire week. Since release windows must align with staffing schedules and business flows, the full week is rarely available, which means that release windows that once afforded six or eight hours now have shrunk to less than an hour. That’s fast.
4) Hybrid Architectures
Today’s operational environments increasingly comprise a mix of on-premise data centers and cloud-based services or apps. More and more on-premise servers are virtualized, leading to server sprawl and its attendant operator overconsumption. In other words, server sprawl leads to the fragmentation of operator attention, as each operator must attend to a spiraling number of virtual servers, a trend exacerbated by the multiplicity of server types involved with modern web apps. Between virtualized application servers, web servers and database servers, data center operators are busier than ever.
Although Unix, Linux and Windows servers dominate at the OS level, many transaction-heavy shops rely on the unparalleled reliability and horsepower of mainframes. Thus today’s operations management regimes must encompass a panoply of operational environments, especially since any given app or service may have dependencies on several of them.
5) Consumerization of IT
End users access IT services on their own phones, tablets and browsers, often while working from home or on the road. This makes IT-user engagement more important than ever. These “IT consumers” expect their IT engagement to travel with them and be presented in the manner they’ve come to expect from contemporary online services. Consider their frame of reference. They reserve flights and get boarding passes on their phone, make restaurant reservations using OpenTable, pay their bills when 3,000 miles from home and even tell their DVR to record the big game when they are nowhere near the TV remote control.
They expect to interact with IT in a similarly handy manner. Fall short and they will turn elsewhere.
Trends happen and responses to them vary, often marking the difference between successful online businesses and those that fall behind or fail entirely. Here are some strategic responses to the IT operations trends identified above.
1) Synchronize People with Process
The entire operations staff should organize around a single process management system. This doesn’t need to occur in one phase, however. For instance, it may start with release management processes or core ITIL processes such as incident, problem and change management. In all cases, the ops orchestration system should enable lean operations. That means it should support pull tickets and provide a visual management system of queue sizes at each stage of operational value chains. Ideally the ops orchestration system can then sync with the system used by dev, forming an IT-wide orchestration system. One benefit of this approach is bridging the so-called DevOps divide, especially in the area of release management.
2) Systematize Demand Management
Use a unified demand management system to pull work through IT and to unify ops with dev. Building on the orchestration system described above, a unified demand management system provides a common request center for IT staff and end users alike. Behind that common front end should be mechanisms to triage the flow of requests, dispatching them appropriately to analysts, developers, end-user support, data center operations, outsourcers and so on.
Specific to ops, such a demand management system provides a consistent response mechanism for 100 percent of the work undertaken by sys admins and operators. It also gives IT a web-mobile face that IT consumers can instantly understand and employ and that is directly engaged with IT’s back-end processes.
3) Automate, Automate, Automate
Automate all repetitive, time-critical tasks. For example, many sys admin tasks remain manual—a condition that carries considerable costs, not all of them visible. Among the invisible costs are quality issues that creep in when fat fingers make a mistake or critical steps are missed. Another is wasted time, a prime impediment to IT operations.
Prioritizing which tasks to automate first should focus on those that impede the agility required for the accelerated cadence described above. A prime candidate for this is application release activities. Even if the data center is increasingly in the cloud, the man-hours required to put a sped-up cadence of web-based app releases into production can be shocking. And it can be automated.
4) Integrate People and System Processes
Integrate people processes, like those in the service desk, with system processes, like application deployment. This requires that the system that automates people processes be capable of orchestrating domain automation systems like release automation and also artifact management systems such as source and deployment code control.
5) Rally around an Operations Calendar
The best ops organizations run crisply: well scheduled and timely. Thus an operations calendar is a perfect visual control system for every part of ops to rally around. Such an ops calendar should be more than a static representation of planned dates; rather, it should be a live system, connected to the ITSM system, the release planning and control system, the monitoring systems, and so on. As a live system, the ops calendar should take multiple forms: traditional calendar, Gantt chart, dependency tree and so on.
In this way, the live ops calendar becomes a single source of truth for everything significant that is happening or planned to happen in ops.
6) Orchestrate Service Management
Service management is the most visible component of IT ops and also the one that is the most process mature owing to ITIL. ITSM, however, remains stuck in the 20th century in too many shops. Orchestrated service management is process based, focused on moving work to conclusion instead of mere record keeping, and it is also visually configurable to keep up with ongoing process changes.
Orchestrated ops is a concept that builds on the tools in place now, using them in a process-driven human workflow. Not only does this approach avoid costly rip-and-replace, it allows for any of the above responses to serve as a starting point.
Pick one, implement it, gain value and then judge which one to pursue next. This is consistent with today’s fast cadence and lean methodology. Incremental gains, consistently achieved, soon add up to a crushing competitive advantage.
The time to get started is now.
About the Author
David Hurwitz, SVP, Serena Software, began his career as a software engineer for legendary Silicon Valley company ASK Computer Systems, the pioneering enterprise applications company. David has spent the 25 years since around enterprise IT, helping businesses benefit from it and helping IT leaders be more effective at it.
Hurwitz holds a B.S. in industrial engineering from Rochester Institute of Technology. His senior project applied animated simulation to industrial robotics for more-effective manufacturing results.
Photo courtesy of IntelFreePress.