Every computer systems in use has operational and development functions associated with it. In enterprises with their own data centres, these are often separated between those that keep the systems running (operators) and those that write applications (developers). DevOps is a term used to describe a software development methodology which combines software development with IT operations. It is different from, but associated with, other terms such as Agile project management, ArchOps, continuous delivery, DataOps, DevSecOps, site reliability engineering, systems administration and WinOps. The challenge for modern IT organisational structures is how to get these two teams to work on the same side rather than in conflict with each other.
Of course it’s not necessary for either of these groups to be employed by the organisation itself and there have always been a wide choice of outsourcing, managed services, packaged software, custom software and other resources which can be brought in to complete IT solutions. Over the last few years these choices have expanded through the addition of SaaS, IaaS, PaaS, BPaaS and other cloud services.
Building an on-premise IT infrastructure involves a number of choices in types of server, operating system and virtualisation software to deploy. Over time strong links have developed at an industry sector level – IBM z System mainframes in the Financial Services industry for transaction processing, Unix and Linux for industrial and high-performance computing, VMware hypervisors for x86 servers for instance (see my Figure opposite). There have been many challenges for the developers who use on-premise systems such as the weeks it takes to deploy new hardware and the different languages needed to programme them. Many of these have been alleviated by the introduction of various cloud services; Infrastructure as a Service offers a complete platform to develop applications on, Platform as a Service offers virtual machines to configure (a function which spreads into IT operational territory).
However the addition of these new resources only make the job harder for IT operators, who have to integrate, manage and protect the resultant applications. When you add Software as a Service (often a complete solution provided to a specific department, such as salesforce.com to sales and marketing for instance), the demands of securing the environment alongside the legal requirements of GDPR and other regulations, it has become almost impossible to provide a holistic view of their organisation’s complete IT landscape.
Almost all enterprise IT vendors provide some tools to help. In particular:
- Microsoft/GitHub – Microsoft’s acquisition of GitHub for $7.5b was announced June 2018 and completed in October. GitHub claims 31m developers and 2.1m business users including airbnb, SAP, IBM, Google, PayPal, Bloonberg, Spotify, Facebook, Node8, NASA, Walmart. Its tools include collaboration, project management and marketplace. It had 800 employees. In Feb 2018 it was victim of a major DDoS attack. Its revenues were c, $140m in 2016. Microsoft of course already had a number of tools including Release Management, Application Insights. Its .Net framework allowed its customers to adopot DevOps strategies, while its Azure cloud offers a commoditised infrastructure to develop on
- Micro Focus – Micro Focus the largest UK-based enterprise software supplier with revenues of $4.6b and 12k employees in the year to the end of Sept 2018. It grew massively in 2017 when it took over most of HPE’s software portfolio. Balancing its books has proven problematic since then and it has been raising money by selling of parts of its business, such as its SuSe Linux division for $2.5b at the end of 2018. Its associated DevOps products, services and solutions are:
- Enterprise DevOps Services
- Application Development, Test, and Delivery
- Micro Focus DevOps Suite
- Micro Focus DevOps Model Office
As per IBM and CA, Micro Focus also supports mainframe and COBOL.
- IBM Rational – IBM had revenues of $80b and 380k employees in the year to the end of Sept 2018. Software is a major contributor to its success, accounting for $24b during that period. Rational was acquired by IBM in 2003 for $2.1b. It was founded in 1981 to expand modern engineering practices – particularly explicit modular architecture and iterative development. It had revenues of $850m and 4k employees before acquisition. IBM’s other DevOps software/services include BlueMix for building applications on the IBM Cloud. It obviously supports DevOps in its z Systems mainframe environment, including applications written in COBOL. It also offers UrbanCode Deploy, which is ‘an application release automation solution that combines robust visibility, traceability, and auditing capabilities’. IBM environments also include Power servers, which run AIX, iOs and Linux, making it core complex than other vendors for DevOps. Its acquisition of Red Hat (see below) enhances its participation in the DevOps world.
- Altassian – an Australian software company with revenues of $874m in the year to June 2018 and 2.7k employees. It claims that its offerings are in use in 125k companies including Airbnb, NASA, Cisco, eBay, Redfin, Toyota and Kaisr Pemanente. Its products ares plit between ‘Enterprise’ and ‘Marketplace’ categories. Specific offerings include:
- Plan, track, and support – JIRA Software, Service and Ops
- Code, build, and ship – Bitbucket, Sourcetree, Bamboo
- Collaborate – Confluence, Trello
Atlassian’s JIRA offerings can also be used in mainframe environments.
- Oracle – had revenues of $39b in 2018 and a headcount of 141k. It is similar to IBM in continuing to produce servers based on its own technology – Sparc systems running Solaris and Linux operating systems (which are also cross-licensed by Fujitsu). It takes responsibility for the continuous development of solutions based on its own resources. Its approach to DevOps is thereby non-generic. As the owner of Java since its acquisition of Sun in 2004 and its widely-used databases it is a strong player in the software development market across all IT infrastructure markets. Like IBM Oracle has become a significant provider of cloud services, which has changed is focus away from mainly on-premise solutions. Oracle is a strong supporter of Open Source DevOp components , which include:
- Configuration Management – Ansible (owned by RedHat since 2015), Puppet, Chef and SaltStack
- Continuous Integration/Continuous Delivery – Ansible, Jenkins/Hudson, Rundeck
- Version Control – Git, Perforce, Subversion
- Virtualization – Docker, Vagrant
- Build and Functional Testing – Gradle, Jenkins/Hudson, Robot
- Binary Management – Artifactory, Nexus
Both its Java Cloud Service and Database Cloud Service include the Oracle Developer Cloud Service, which are used by over 21 product development groups within Oracle itself. This includes simplified development, collaboration and management, automatic deployment and integration with JDeveloper, Eclipse and NetBeans.
- Broadcom/CA – CA had revenues of $4b and 11.6k employees in the year to Sept 2018. It was surprisingly acquired by Broadcom for $18.9b in November 2018. Its products are split between enterprise and mainframe solutions (alongside services). Its software supports Agile project management. Its product categories are:
- “Plan – Get the agility you need to drive change—from the boardroom to development.
- Build – Ignite innovation by continuously developing quality apps faster and more efficiently.
- Test – Continuous testing helps you build better apps, faster.
- Release – Accelerate and optimize app releases for the modern software factory.
- Operate – Maximize visibility from mobile to the mainframe so you can maximize your customer’s experience.
- Secure – Secure your digital enterprise. Prevent breaches. Provide a seamless customer experience.”
Broadcom cited the importance of a stake in the mainframe market as a partial rationale for the acquisition; however there are virtually no links with Broadcom’s chip manufacturing business.
- Red Hat – revenues of $3.2b in the year to the end of August, it is one of the most important vendors for enterprises seeking to use Open Source software components in their IT solutions. As the owner of Ansible it is a direct supplier of DevOps tools. It was acquired by IBM in 2018.
- HPE – revenues of $30b in the year to October 2018. Despite selling off the majority of its software to Micro Focus in 2017 (see above) its OneView IT infrastructure management product is essential for its large customers who want to create pools of flexible resource.
These are just a few examples of course. As cloud computing develops it’s important not to forget about those running and building apps on in-house systems. It’s interesting that these appear in IBM’s cloud computing model (see my Figure below), even if relegated to a small section to the bottom right.
DevOps has to accommodate any IT infrastructure element of course and things got more complicated from 2002 onwards with the rise in the use of Vmware and other hypervisors on x86 servers… as it has with the rise of Docker container applications. The number of tools available expanded significantly with the growing use of Linux and importance of Open Source software projects.
Things arguably got easier for end-users as the large cloud suppliers (Google, Amazon, Microsoft, IBM, etc.) built out their businesses. For cloud services (IaaS and PaaS) to succeed they have to provide easy-to-use ‘composable’ services and applications. They each provide a single set of tools for DevOps, cutting down the multiple choice users have when building their own on-premise solutions. 2018 saw the beginning of ‘multi-cloud’ strategies for those companies who want to straddle the use of on-premise computing with cloud services from outside.
The success of cloud approaches have now led many running on-premise platforms to seek out simpler methodologies and processes. Using RESTful APIs allow management systems to work together, while Open Source software has reduced costs for those with good levels of technical skill.
In the future AI will help to reduce the multiple manual processes that many have to deploy in IT operations, as well as those to integrate code by developers. Nevertheless to get the most from DevOps many organisations still have to change the way the run their IT – too many still compartmentalise processes, in advertently sustaining a civil war between those developing applications and those keeping the lights on.
©ITCandor Limited – unauthorised copying of this content is illegal and will be rigorously defended by us through court action