IT pros contend with the future of continuous delivery
Continuous delivery tools are poised to add comprehensive breadth and analytical depth, but whether most enterprises can assimilate them is another matter.
SAN FRANCISCO -- Just as mainstream enterprises get comfortable with continuous delivery, another wave of change is gathering force.
CloudBees' forthcoming software delivery management (SDM) platform will tie together several fresh trends for continuous delivery tools, from analytics-based observability to bleeding-edge workflow concepts such as BizDevOps. The ultimate goal is progressive delivery, a stage of maturity that lies beyond current efforts to shift IT processes to the early parts of the software development processes and fully automate testing, infrastructure management and application deployment. Under progressive delivery, iterative changes to code are pushed to production incrementally with methods such as canary deployments and feature flags.
But every fresh innovation in continuous delivery requires its own kind of organizational regression testing within enterprise IT shops, and often painful backporting to legacy infrastructure, such as mainframes, that is ill-suited to cloud-native notions of IT management.
At global financial services firm HSBC, engineers must work with CloudBees to adapt the enterprise version of Jenkins, CloudBees Core, as well as SDM, to mainframe and iSeries midrange systems that Jenkins doesn't support today.
"We must include mainframe and iSeries [in continuous delivery] in order to achieve full-scale agility and speed," said Rajeev Mahajan, CIO of DevOps at the London-based enterprise. Mahajan estimated that IBM z Systems mainframe and iSeries midrange constitute more than half of HSBC's 90,000 servers worldwide.
"We don't want DevOps to be selective," Mahajan said. "We want a uniform CI/CD experience."
So far, HSBC itself has painstakingly integrated mainframe and midrange application test automation and code deployment tools into homegrown continuous delivery pipelines. The firm has also built some of its own continuous delivery automation tools, such as the Shunya engineering management framework and an AI-based defect analysis tool called SamurAI, and wants to push the envelope on automated, analytics-driven pipelines to production.
Rajeev Mahajan, CIO of DevOps engineering at HSBC, talks about his company's DevOps transformation at DevOps World | Jenkins World.
HSBC engineers collaborate with CloudBees on packaged mainframe pipelines that they hope to integrate with SDM, which will offer a much broader set of capabilities than HSBC has been able to achieve on its own, Mahajan said in a DevOps World | Jenkins World conference session here this week.
"It's along the same lines [as Shunya and SamurAI], but SDM takes a very holistic approach to all the data that is being generated by all the [DevOps governance] rules," Mahajan said. "CloudBees sees all the technology organizations and governs the full spectrum [of stakeholders], while we're only doing this for our line of business."
IT pros struggle toward continuous delivery maturity
Even as it pushes to integrate legacy systems, HSBC is light years ahead of most enterprise IT organizations in its ability to absorb advanced continuous delivery tools. Other shops must still get a handle on data-driven IT automation, and aren't sure yet how existing efforts based on third-party log analytics and monitoring tools such as the Elastic Stack will fit into SDM.
It's been an ongoing open question: Why does the CIO not have the degree of automation that his or her peers have in HR, finance and manufacturing operations? Why do SAP and Oracle not have a module for the CIO?
Charles BetzAnalyst, Forrester Research
"Everyone is on board with the idea of some kind of universal data aggregation layer, and our Elastic Stack deployment consolidated a lot of different efforts," said a solutions architect for an insurance company on the East Coast who requested anonymity. "But it's not just the tool, but the process behind it that's been the challenge, getting access to information across an organization like ours."
Consolidating all data-gathering tools under the SDM umbrella would be helpful, but that will depend on how well-integrated those existing tools are, which remains to be seen in the product's early stages, he said.
In some ways, the IT industry has long anticipated continuous delivery's future capabilities, particularly in data-driven automation. But the struggle to make a workable ERP-like workflow automation system for IT is longstanding, too, analysts said.
"On the main stage, [CloudBees] didn't mention SAP, but the concept of ERP for IT has been around for at least 17 years," said Charles Betz, analyst at Forrester Research. "Since then it's been an ongoing open question: Why does the CIO not have the degree of automation that his or her peers have in HR, finance and manufacturing operations? Why do SAP and Oracle not have a module for the CIO?"
IT vendors such as BMC and ServiceNow have also attempted ERP-for-IT tools, and it's tempting to be skeptical as CloudBees SDM appears headed down the same path, Betz said. But SDM's CI/CD-based view on data-driven DevOps processes, as opposed to IT operations systems, might set it apart from those earlier efforts.
"The reason ERP for IT has struggled is that [previous versions] have been operational platforms," Betz said. "CloudBees has a focus on the actual building of software products -- the critical parts of the process CIOs more want visibility into, from development to deployment."
