IBM Destination Z - Group home

Mainframes in the Cloud

By Destination Z posted Mon December 23, 2019 03:27 PM

The cloud is what people who know about these things call a nominalization. In Neuro-linguistic programming terms, it’s something that can’t be put into a wheelbarrow—i.e., it doesn’t physically exist! And yet, we spend much of our time uploading stuff to it and using it.

In the hands of the marketing people, it’s a service that we buy/use—i.e., software, platforms, infrastructure. Basically, the idea is that whatever we need exists somewhere, and we don’t need to worry about any aspect of it. The provider has to ensure that it’s online and available—or else we can get the cloud service from somewhere else.

So, who are the big players in cloud services? Amazon, Google, Microsoft Azure, Oracle Cloud, Rackspace, Salesforce, Zoho and, of course, IBM. The basic architecture is one where the provider runs a number of servers, and these are virtually subdivided into as many virtual servers as possible to run the systems at speeds that users want. The basic architecture is one where the provider runs a number of servers, and these are virtually subdivided into as many virtual servers as possible to run the systems at speeds that users want. The next question is this: Which company do you know of that has been able to virtualize operating systems on one set of hardware for a very long time? Need a clue? They invented the VM hypervisor back in the 1960s (i.e., CP-67).

As we all know, IBM mainframes can be subdivided to run a number of different partitions, and those partitions can run z/OS, z/VM, z/VSE and z/Linux—or to give it its official title, Linux on System z. Users can install Red Hat or SUSE Linux. SUSE was acquired by Novell in 2003, before Novell itself was acquired by Attachmate in 2011. The net result of running Linux on a mainframe is that large companies can take all their numerous Linux servers and virtual servers and put them on a single mainframe—and enjoy all the benefits of uptime, security, etc., that we know go with using mainframes.

So the big question is: Why are only a few sites taking up this option? What’s the sticking point for so many others, who must be aware of the opportunity? The answer, when you chat to people, seems to be mainly price. And that’s what a large part of IBM’s announcement last month (during the 50th birthday celebrations, read it here) was designed to address. The other part was to provide rapid deployment.

IBM claims that their new IBM Enterprise Cloud System “provides an integrated platform, built upon open standards, for clients and service providers looking to rapidly build out a trusted cloud environment capable of supporting mission-critical workloads.” They also announced a new flexible utility pricing model that gives service providers the ability to pay for Linux-based mainframe cloud infrastructure over time, based on compute consumption, rather than system capacity.

IBM’s unique selling point for mainframes in the cloud is their ability to support up to 6,000 virtual machines in a single system, provide a secure multi-tenant environment and dynamically share resources across workloads. IBM suggests that with higher system efficiency and greater scalability, the total cost of some Linux on System z cloud deployments can by up to 55 percent less than comparable x86-based cloud infrastructure.

The IBM Enterprise Cloud System is factory built and configured with automated cloud orchestration and monitoring to allow purchasers to rapidly deploy enterprise-grade cloud services. It provides an Infrastructure as a Service solution. Customers can make use of specialty processors, particularly IFL, which was introduced in 2000. So, a zEC12 can have 100 IFLs. Each IFL can host 60 virtual machines—giving the total of 6,000 virtual machines for a single mainframe host.

The Enterprise Cloud System also comes with IBM v7000 or DS8000 storage, IBM Wave z/VM (a graphical management interface for managing VMs), and IBM Cloud Management Suite. The Cloud Management Suite includes SmartCloud Orchestrator for configuring and deploying virtual machines, Omegamon XE for monitoring and managing performance of workloads and Tivoli Storage Manager. This is what makes setting up, using and managing Linux virtual machines so straightforward.

It will be very interesting to see how much take up there is for this and what IBM might be prepared to offer next, in order to get a larger portion of what is going to be a very lucrative market.