“Mainframe” is the term most people associate with the ultra-scalable and ultra-reliable Enterprise Servers that many large enterprises run their core business on.
Enterprise Servers have moved through three phases. First, specialist proprietary boxes (such as the 390 Mainframe), where you paid for potential capacity and service levels you might not actually need, in return for unparalleled reliability, capacity, availability and security (these Enterprise Servers – Mainframes – are still widely used in the biggest companies). Second, distributed clusters of Intel-based commodity servers, cheap if service levels and resilience aren’t important (less cheap if they are), culminating in server sprawl and expensively-managed server farms (these, also, are still in use). Third, we have Enterprise Server 3.0 – ES-3 – which is also called Mainframe 3.0, to emphasize continuity with what came before.
ES-3 is server consolidation: with the “ultimate cloud in a box” – or Infrastructure as a Service (IaaS) – and this is, in part, driven by a need to control the electrical power consumption (and heat dissipation) associated with enterprise business automation; and to achieve economies of scale. The benefits of ES-3 are enabled by the secure multi-tenanting and rock-solid virtualisation which architectures such as IBM z are proven to deliver (and have delivered, in previous incarnations, before products like VMWare were invented).
Enterprise Server 3.0 is not (just) hardware in a box. It is hardware complemented with management software for workload scheduling, throughput optimisation, resource usage optimisation and resilience – it is a combination of hardware and software delivering a bullet-proof infrastructure as a service to the enterprise.
Technologies and capabilities important to ES-3 (although they are frequently implemented on other platforms too) include:
- Transaction processing monitors and transaction processing DBMS;
- Workload scheduling and operational analytics and management;
- Technology virtualisation as a service and Agile Storage;
- DevOps at scale.
ES-3, to a large extent, represents “Mainframe culture”, which is risk adverse and oriented towards preserving critical business service levels as technology evolves and business needs and workload patterns change – without the luxury of “planned downtime”. Cultural integration with “everything runs on beta” cultures can be a bigger issue than technology integration. An organisation needs both cultures, risk-averse where it matters, but risk-embracing and innovative where this risk can be managed.
ES-3 arose from the “grave of the Mainframe” in 2011. ES-3 is, in essence, the Mainframe without its traditional limitations (it has fast CPUs and can run distributed Unix and Windows applications natively on Intel blades). It is services- and standards-oriented; with ES3, enterprise automation run on virtual Linux machines using open-standards protocols, sharing a server which delivers less than 5 minutes downtime a year; mixed workloads are scheduled automatically for maximum throughput at the largest scale (with IBM’s Unified Resource Manager, for example); and hundreds of virtual machines can run in parallel on fast CPUs, with absolute isolation. Using ES-3 cloud services, users can provision resources on demand and de-provision them as soon as they are no longer needed. Freeing them up for other users – and only paying for what they actually use.
The ES-3 machine of the future is going to be a hybrid cloud appliance, supporting an in-house private or (increasingly) a public cloud, as well as providing a traditional enterprise computing platform, where appropriate. It is powered by an architecture designed from the ground up to support multi-processing, multi-tenancy, high availability and high security; with extreme energy efficiency. It will be bought by large enterprises seeking to rebuild their internal IT out of cloud services and by the new breed of SI, the Service Integrator, selling business automation built out of orchestrated cloud services.
In the meantime, the widespread use of Mainframes in the largest enterprises supports a flourishing ecosystem, albeit not one terribly visible to general IT. This supports, for example:
- Development and maintenance of Mainframe applications with a special focus on cross-platform development (for example, developing/maintaining mainframe zOS CICS COBOL applications on Windows);
- Mainframe software asset management;
- Migration of applications between Mainframe and distributed platforms (including legacy modernisation);
- Systems and operations management, including Mainframe operational analytics and Mainframe test data management;
- Mainframe session and workload management;
- Mainframe security and access control;
- Data centre automation, including Mainframe printing and RFID processing;
- Mainframe middleware for integrating Mainframe and ES-3 systems with corporate distributed systems, including web-enabling Mainframe OLTP (eg CICS) applications, publishing Mainframe output as XML/HTML, desktop access to Mainframe data, and the use of secure mainframe features for, for example, Web-commerce credit card processing, encryption and identity management.
- Outsourcing of Mainframe services and support and managed services for the Mainframe.
- Mainframe DBMSs
Any enterprise seeking to modernise a mainframe legacy should look at ES-3, because it delivers the same workload throughput, performance, resilience and security that made the Mainframe the centre of the enterprise – but more economically and more flexibly. It delivers value from reuse of the software the enterprise already depends on and from innovative knowledge-based management tools.
Any enterprise wanting to control “server sprawl” without impacting service levels should look at ES-3, because IBM (for example) provides documented use cases of it doing just that (see here, for example). ES-3 delivers value in space-efficient automation.
Any enterprise with “green computing” energy efficiency issues or which is hitting datacentre electricity supply issues should look at ES-3, because it manages heat more efficiently than most server farms and runs at 80-100% utilisation – an order of magnitude (at least) higher than distributed servers in general. ES-3 delivers value in energy-efficient automation.
Any regulated enterprise or one that holds personal data should consider ES-3 from the point-of-view of its resilience and bullet-proof security (especially as a basis for a cloud security model).
Developers who are frightened of legacy Mainframe development (this is largely a perception issue, as the logic of programming is much the same everywhere and even distributed systems now have the virtualisation and parallel processing capabilities once only available on Mainframes) now see an ES-3 opportunity for the incorporation of cost-effective mainframe services (hardware-based encryption services are a notable example) in business systems. Even younger developers are now noticing that new ES-3 tools are removing the barriers (nomenclature and user interface) that once made moving into a Mainframe environment difficult.
A start-up that wants to be able to grow without limit without over-provisioning in the early stages should look at ES-3, because of its services model and because vendors like IBM will move in extra capacity in advance of your need; and only charge when you start using it. ES-3 can deliver value from a cloud-based infrastructure agility model, as well as delivering value from a conventional on-premise server model.
ES-3 computing is all about running workloads on the most appropriate platform for that workload – sometimes this will be ES-3, sometimes it won’t.
The ES-3 marketplace is dominated by IBM and z Enterprise, with companies like BMC, Broadcom and Rocket Software (as well as IBM) delivering its management software. However, ES-3 is not one particular technology and any enterprise server technology could qualify if it can match z Enterprise levels of scalability, resilience, availability, security, energy-efficiency and so on. Provided, that is, it can do this for long-term processing of real workloads without “planned downtime” let-out clauses and so on.
There are examples of such machines from several vendors (especially in Japan); possibly, some of these are not moving on significantly from the legacy “mainframe” concept. The IBM z enterprise servers, however, with built-in distributed blade-servers (a hybrid architecture), fast CPU’s and more sensible pricing models are the exemplars of ES-3. Key features are:
- Efficiency at scale, with a hybrid architecture supporting native running of mixed workloads.
- Strong support for parallelisation (e.g. with transactional memory), resilient virtualisation and cloud deployment.
- Extreme resilience and trusted security.
- Accessible server management tools.
Emerging trends are the introduction of more customer-friendly pricing models (with special new-customer incentives) and the availability of better management utilities. These new utilities often have built-in knowledge transfer features, to allow distributed-computing support technicians to pick up ES-3 support skills rapidly, and integrate management across all technology platforms from a single point of control.
Increasingly, management will be based on sophisticated, near-real-time operational analytics (e.g. zAware), running in its own secure partition.
Distributed technology vendors are producing Intel-based Enterprise servers that claim to emulate ES-3 service levels – although IBM’s z architecture probably has a more established record for actually delivering these service levels.
Some way out, ES-3 is likely to find a home powering public enterprise cloud services from an efficient multi-tenanted ES-3 host.
DevOps, the extension of Agile processes to Operations, and collaboration across Dev and Ops silos, supported by automation, will help to bring ES-3 platforms into the IT mainstream.
Finally, as usual for mature marketplaces, consolidation into large vendor offerings is continuing.
Hardware vendors supporting the ES-3 space are consolidating, and vendors such as Hitachi have effectively left this space to IBM, although IBM probably doesn’t want a monopoly position (for regulatory reasons) and there is a continuing space for software vendors that don’t have a hardware business to support. Nevertheless, vendors such as Unisys still take an independent approach to ES-3 and, arguably, new servers, such as those in Amazon’s Arm-based Graviton processor family, are ES-3 class.
Since so much of Enterprise business still runs on ES-3, there is still a flourishing ES-3 ecosystem, although there is also a trend towards smaller companies in this space being acquired by bigger players and there is increasing consolidation of software tools, with Broadcom acquiring CA technologies, BMC acquiring Compuware, and Rocket acquiring ASG and Uniface, for example. One interesting sign of the times is that Compuware, which divested itself of a lot of interesting technologies as new companies such as Uniface, Changepoint, Dynatrace etc, didn’t just sit back and milk its locked-in Mainframe customers (as many expected) but actively innovated in the Mainframe space, and has been acquired by BMC, which is actively supporting its innovations (especially in ES-3 DevOps).
Areas where the vendor landscape is not changing, are of some interest – databases such as Adabas, Nomad and Focus; and tools such as Syncsort; still have a life (although who actually owns them may be more mutable).
Bloor expects to see a new class of emerging cloud-services integrators emerge, using technology to provide higher-level enterprise services, although when this will happen is moot.