Andrew Price: All longtime NonStop users have watched over the years as the NonStop platform has become more and more standards-based. From the point of view of a NonStop end-user (developer and/or administrator), what do you think have been the most beneficial moves in standards adoption?
Franz Koenig: With the seven domains of modernization that I had defined many years ago, the goal was not only to bring newer technologies, but also to adopt many more industry standards, pseudo standards and practices. These seven domains we focused on were modern user interfaces; open, service oriented integrations; state of the art information layer; open service run-time environments; modern development tools and processes; security and last but not least the manageability aspect. Today end users can enjoy the same modern web- or mobile user interfaces using standard frameworks, but also easily expose their service and data assets quite easily. Application development for NonStop is like for any other platform, i.e. companies can leverage ubiquitous skills, as development tools, APIs, frameworks, app-servers, programming languages, but also deployment tools are just like on other platforms. Access to our SQL/MX database is standards based, and with tools like Hibernate or JPA applications can be moved from other platforms to NonStop without code changes. Customers or partners who want to bring new solutions to NonStop that were designed for OracleTM , our new PL/MX support will significantly accelerate and ease such efforts, and with multi-tenant Database Service (NonStop DBS) and NonStop Database Analyzer we boost manageability. Talking about manageability – the work we do around DevOps, virtualized NonStop (vNS) and industry standard automation management (e.g. through Ansible) gives customers more flexibility, choice and power than ever to easily provision and manage NonStop infrastructures.
Price: What architectural/software design trends would you like NonStop developers to know about right now?
Koenig: Driven by an amazing wave of digitization and industry shifts through disruptive innovations and new business models, enterprises need a super robust backbone that provides them with both the architectural quality like scalability and availability, but also the foundation to create new services faster than their competitors using state of the art technologies, We see a lot of interest in microservices and reactive design, API driven economy (REST APIs), hybrid analytical transactional data stores, robust in-memory data stores, cloud native design principles and most certainly DevOps.
Following all the trends may be really hard for customers. This is why my team, the Advanced Technology Center (ATC), can assist customers all the way from desired business outcomes to technical architecture and design ensuring they get what they need to win in their businesses. We do leverage various methodologies ranging from strategy to architecture to design thinking.
Price: Do you have a view to how much new Java code is being written for NonStop applications, versus other languages?
Koenig: For new services and applications we see the growing adoption of the Java ecosystem. The reason I call it “ecosystem” is in the fact that in recent years people have not only been using the Java language in the Java Virtual Machine (JVM), but several other languages often referred to as JVM, such as Scala, Groovy, Jython or Kotlin became quite popular. These languages are all translated into Java Byte Code just like the Java language itself, and so these new languages cannot only be easily combined with each other or Java, but also take full advantage of the power of the JVM.
NonStop’s Java portfolio is quite comprehensive, with the NSJSP (Tomcat based) Web Container, the JBoss Wildfly based NSASJ, the ActiveMQ based message queuing product NSMQ, plus a magnitude of popular frameworks/open source tools and products. With that we make it super easy for software engineers, who never wrote a program for the NonStop, to be productive immediately, giving companies the advantage to leverage ubiquitous programming skills with the NonStop while still benefiting from the platform fundamentals.
Price: Has Nonstop gotten past the perception (reality?) that Java on NonStop has performance issues?
Koenig: Java performance on NonStop is very good for most business applications. We’ve just successfully benchmarked a new solution at 5000+ business-transactions per second at great response times. People had been complaining about Java performance on the NonStop in the past. This was mostly caused by atypical tests (e.g. pure number crunching), and often times we’d been pulled in for business apps/services performance problems as well. It turned out that the root cause in almost every case was around significant I/O inefficiencies. NonStop’s I/O architecture is less forgiving to inefficiencies, such as excessive logging/file-open calls, wrong transaction management or simply inefficient I/Os (e.g. byte-wise file I/Os. People do compare the NonStop performance with workstation performance, often not realizing that they’d not be able to sustain higher transaction loads at any cost on such systems, even when throwing massive amounts of hardware against the problem. When these bad practices are resolved, NonStop Java performance is really good and competitive. And we are working hard to make it even better for additional new workloads – so watch the space!
Price: Do you see technologies such as REST having a significant impact on NonStop applications? How?
Koenig: REST certainly is prevalent and still growing in popularity. REST has all the attributes of other Internet successes – it has a very clear value and simplicity ultimately always wins. Being around in this industry since 34 years I have seen this over and over again. When I started my career OSI with its seven layers was considered state of the art, super powerful network stack with specifications spanning multiple books. OSI was considered much superior to TCP/IP, but eventually failed due to its complexity to implement and manage – and we all witness the continued success of TCP/IP. Same with service orientation: SOAP was considered much more powerful and superior, but REST won.
Not only do I see REST to be highly attractive for exposing key enterprise assets like services or data, but also in areas like manageability. A wide range of offerings is available for the NonStop through ISV Partners, Open Source and HPE ranging from API Gateways, Microservice and other frameworks, Libraries, Application Servers, etc. Some partner offerings are even available for Guardian.
Price: What has been the most exciting project that you and your ATC staff have worked on?
Koenig: We have been working on so many exciting projects over the many years, it is really hard to pick the most exciting one. Our focus is around three value chains and lifecycles – customer, partner and products. The most exciting customer engagements are those that allow us to have significant impact on the customer’s desired business outcomes. On the products side it is very exciting to see innovations triggered and incubated by the ATC succeed and add significant value for our customers and partners. Last, but not least, helping partners to improve their product, e.g. by analyzing and boosting solution performance and see them win in the market is a really great
Price: What current NonStop capability are you most proud of?
Koenig: It may sound boring to some of the longtime friends of NonStop, but the topic is more relevant than ever. NonStop’s architectural principles – to scale-out rather than scale-up, the shared nothing architecture, the message based design combined with the unmatched availability and data integrity. If you look at today’s cloud architectures you can see a big shift in mindsets. While in the past the industry favored large, scale-up SMP style machines, these new architectures are all about scale-out and stateless services. However, real business applications require you to store state safely and consistently somewhere, and this is where cloud implementations delegate the challenge to the application developer, while NonStop does all the hard parts for you already, no matter if you chose converged or virtual NonStop. This is why NonStop is more relevant than ever and our virtualization, cloud and “as-a-service” efforts, such as NonStop Database Services, the fault tolerant in-memory database (NSIMC), the unique way we can support fault-tolerant stateful services, give our customers unique value!
Price: What technologies are you watching for potential future NonStop support?
Koenig: Having a background in innovation management, design thinking and strategy I do apply many of the best practices from these domains. I maintain a technology radar, verify what unmet customer needs or pains a particular technology can close, see if the timing is right, verify its potential business impact, possible business models, go-to market options, technical fit and feasibility, etc. In many cases we pick promising candidate technologies and perform time-boxed evaluations either in the ATC only, or as a larger effort that includes engineering and product management. This approach allows us to filter out and focus on the really relevant technologies and turn some of these into NonStop (or partner) products/-features.
Good examples of this approach are NonStop DBS and the forthcoming NonStop Middleware Manageability based on Ansible. NonStop Database Services (DBS) turns SQL/MX into a unique multi-tenant database as a service offering. Watching market needs and technology trends I had this on my radar for a while. After I conceptualized an architecture model, Frans Jongma and Cor Geboers (both senior ATC staff) created a fully functional prototype in very short time, which we then turned over to NonStop product management and engineering to productize DBS. Not only is NonStop DBS the basis for HPE-IT’s data architecture, but also a great way for existing and new customers to solve some of the toughest challenges in their digital core in a very elegant and modern way on NonStop X and Virtualized NonStop. NonStop Middleware Manageability arose from my modernization blueprint, for which I had many techniques and technologies on the radar, such as DevOps, Continuous Integration/Deployment (CI/CD) or automated configuration management. In times of Virtualized NonStop the automation management becomes a really critical topic. The open source Ansible tool turned out to be a really great candidate to address this need, such that we’d ultimate get towards single click deployments of virtualized systems including the entire software stacks and business services, but also to make installation/configuration changes to a number of systems in a single operation. Cor Geboers did the initial groundbreaking work by creating Ansible playbooks for several NonStop products, which Meg Watson and Abhijeet Walimbe (all ATC members) leveraged and validated internally and in customer projects. Again, this work and insights have been handed over to product management and engineering and from that the upcoming NonStop Middleware Manageability support emerged.