Powering the Cloud
The Power Systems software stack is built around making it easy for users to deploy and manage an enterprise-grade cloud solution.
By Kristin Lewotsky08/01/2017
In a business climate marked by change, the ability to adapt is essential. And savvy IT professionals know adaptability lies in building an infrastructure that can address everybody’s computing needs quickly, efficiently, securely and cost-effectively.
When it comes to adaptability, infrastructure matters. The IBM Power Systems* platform offers the capacity, blistering speed and rock-solid security required to run mission-critical applications. Hardware is only half of the solution, though. Building a responsive platform requires the right software stack to deliver those computing resources to internal and external users. And for modern enterprises, building a responsive platform also requires the cloud.
“One of the things we’ve paid great attention to as we’ve built our cloud technology is to make sure the solution is production-level quality, just like the rest of our product stack. ... We are allowing clients to get the benefits of the unique features of the platform.”—Carl Burnett, Distinguished Engineer, Power Systems Software development
The promise of cloud infrastructure is one of complete flexibility. Organizations invest in computing capacity on or off premises, then access it as required—no more downtime or resource limitations. It’s an appealing model, designed to satisfy demand while minimizing infrastructure investment. For organizations running mission-critical workloads, however, the public cloud that backs up music and family photos won’t suffice. Enterprise workloads need enterprise-grade cloud infrastructure.
Enter the Power Systems Software* stack, designed top to bottom to make both public and private clouds more flexible, reliable, scalable, and above all, more secure. Leveraging OpenStack technology, the software is crafted for both interoperability, and fast and effective cloud deployment. Because the software stack is built around the Power Systems platform, it also retains the mission-critical capabilities that have always been its hallmark.
The stack consists of the PowerVM* virtualization layer, the PowerVC* virtualization management layer, PowerHA* high availability and PowerSC* security and compliance. With these four tools, IT departments can efficiently manage their cloud architecture—whether that’s on-premises, off-premises or hybrid. Not only can IT shops deliver the services their organizations expect, but they can do it more easily and cost effectively. In addition, the PowerVM and PowerVC offerings are compliant with OpenStack, the open-source cloud OS. This ensures the Power Systems Software stack is familiar and interoperable with other OpenStack cloud management tools.
“The clients who have made the leap to a cloud architecture are beginning to understand the IT savings they can get from those models,” says Carl Burnett, Distinguished Engineer, Power Systems Software development. “They are seeing that it frees up their IT staff to focus on other problems in the environment and not just on managing the infrastructure.”
Virtualization With PowerVM
PowerVM virtualization comes preinstalled on every Power Systems server. It virtualizes the resources available on the machine, including CPU, memory, and I/O such as networking or storage. It enables multiple LPARs to be defined and provisioned on a single server.
Traditionally, PowerVM virtualization has been built around two key software components: the Power Systems hypervisor and the Virtual I/O Server (VIOS). The hypervisor enables CPU and memory resources to be shared among LPARs on the server. VIOS allows the client LPARs within the server to share physical I/O resources. The hypervisor ships with the system, while VIOS is either installed at the factory or by the client (and tailored to their I/O configuration). PowerVM virtualization also provides a high level of isolation and security between partitions.
Into the Cloud With PowerVC
Although virtualization within a server is useful, true flexibility and reliability come with the ability to move and provision workloads across multiple servers. PowerVM virtualization provides this functionality via the HMC. A third major element of PowerVM, the HMC is an optional, externally located hardware or software appliance that manages server resources at a systems level. It’s a more efficient way to define LPARs, and can also be used to create and run automated scripts to simplify system administration tasks.
With the increasing emphasis on cloud infrastructure, system-level tasks have become both more frequent and more complex. In response, IBM introduced PowerVC virtualization management, its OpenStack-based cloud management layer. PowerVC virtualization management brings server, storage and networking resources into a single portal to permit efficient management of all of the assets on the system. The software is regularly updated to stay current with the latest OpenStack releases.
“If a client doesn’t want to use PowerVC, they can manage Power Systems assets through PowerVM NovaLink using a plain vanilla OpenStack-based manager.”—Rajesh Rengarajan, vice president, Power Systems Software
PowerVC virtualization management significantly speeds up the process of provisioning virtual machines. “A client was building out two data centers with POWER,” Burnett says. “In one data center, they used the traditional HMC management model with PowerVM. It took them two weeks to deploy 60 virtual servers. In the other data center, they deployed PowerVC with PowerVM. The software enabled them to deploy those same 60 servers in two hours.”
Able to manage up to 5,000 virtual machines, PowerVC virtualization management features the Dynamic Resource Optimizer (DRO), which can automatically balance workloads across multiple LPARs and servers. The PowerVC DRO can be combined with the Power Systems capacity upgrade on demand (CUoD) CPU and memory features to create a resilient, highly scalable system. For a nominal fee, Power Systems users can buy excess capacity that’s left inactivated—so-called dark cores and memory. When the need arises, the licenses for the cores and memory are activated. Applied with DRO in a virtualized environment, dark cores can be used for instant, affordable scalability.
Consider a big data application in which the system momentarily runs out of resources to process the input. In a typical cloud environment, the workload would be moved to a different server. In a CUoD implementation, DRO would instantly activate the excess capacity on the same server. “It’s a case of moving compute to data as opposed to moving data to compute,” says Chet Mehta, Distinguished Engineer, Power Systems Software development. “A lot of these users are looking for real-time statistics. If you have to move your data to a different host, it might take 20 minutes before you can resume analysis, but you may be able to add some cores to an application in a fraction of a second.”
PowerVC virtualization management has a self-service capability that enables users to request computing resources without assistance from IT staff. The system administrators establish rules and templates for the types of virtual servers that can be deployed. Users log in and, if they have privileges, request a server. One will be provisioned for them automatically and returned to the cloud when they’re finished. Once parameters are set for the process, provisioning takes place without any effort on the part of the IT staff.
Designing for Interoperability
Although the HMC provides significant benefits, it’s not optimized for cloud architectures. As a result, IBM added another element to PowerVM called NovaLink. This virtualization management component is designed to streamline the interface between PowerVM and OpenStack cloud management tools. Administrators can manage the server through services within NovaLink or they can use PowerVC virtualization management.
PowerVM virtualization supports AIX*, IBM i and Linux* LPARs. PowerVC virtualization management runs in a Linux partition, although it can interface with partitions running any of the other OSs.
Given that PowerVC virtualization management is OpenStack-based, it’s also compatible with a number of other cloud orchestrators. “We worked on making sure PowerVC works with cloud orchestrators like vRealize, which is from VMware,” says Rajesh Rengarajan, vice president, Power Systems Software. “Many clients have VMware and vRealize in their environment. Being able to manage assets through vRealize on their x86 boxes as well as their Power Systems hardware, for example, is an important advantage.”
The OpenStack compliance of PowerVM adds another level of flexibility. “If a client doesn’t want to use PowerVC, they can manage Power Systems assets through PowerVM NovaLink using a plain vanilla OpenStack-based manager,” Rengarajan notes. “In this way, we are allowing clients the opportunity to use any OpenStack technology they want to manage PowerVM.”
From financial services to healthcare, retail to government, AIX on the Power Systems platform supports a host of mission-critical workloads. To ensure they keep running, IBM developed PowerHA high availability. The AIX management solution is designed to monitor the operation of computing assets. If the infrastructure or application fails, PowerHA high availability leverages redundant infrastructure to restore the application as quickly as possible.
It’s available in two versions: The Standard Edition manages availability among servers within a given data center, and the Enterprise Edition enables high availability management across different data centers separated by as much as 100 km.
Cloud architecture can offer additional benefits by marrying availability with cost efficiency. Because the PowerHA offering is deeply integrated with PowerVM and Power Systems hardware, it can leverage CUoD for failover operations. Instead of keeping full redundant active cores, memory and related software licenses on a live secondary server, PowerHA high availability can relocate the software licenses for the required CPUs and memory resources on the secondary server, activate them and then restart the application. “The client has a high-availability environment without having to actually pay for twice the resources,” Burnett says. “The system administrator defines policies and PowerHA orchestrates all operations to get the needed resources activated.”
Security and Compliance
Business operates within a complex and detailed regulatory environment. Policies and compliance requirements exist for financial services, healthcare, utilities and more. This extends to security, covering everything from the length of a password to the specific networking ports to be used. Implementing these kinds of procedures on a single server might be straightforward. Doing it on dozens, or thousands, of virtual machines or servers can be arduous. PowerSC security compliance provides a tool for streamlining this process.
It comes with built-in security profiles designed to simplify compliance. It’s essentially a security rule manager that enables clients to get the level of protection they need for the resources on their servers. Profiles can be customized and deployed on AIX in the client’s environment. A new centralized management console simplifies profile deployment across a large number of AIX servers via a GUI.
Optimized for Cloud
Despite the promise of cloud computing, IT shops that are ticking along just fine with workloads using local virtualization may have concerns about switching to a cloud-based deployment. According to Burnett, they shouldn’t. The Power Systems Software stack is built around making it easy for users to deploy and manage an enterprise-grade cloud solution. “One of the things we’ve paid great attention to as we’ve built our cloud technology is to make sure the solution is production-level quality, just like the rest of our product stack,” he says. “We are allowing clients to get the benefits of the unique features of the platform.”
IBM is deeply committed to this effort, just as they are committed to supporting users with compliant open-source products. “The key message is we are leveraging open-source technology and open-source management tooling so there is no skill mismatch between what Power Systems Software requires and what our clients have,” says Mehta. “The core value proposition of the Power Systems architecture and integrated Power Systems Software stack is something that we are continuing to invest in.”
OpenStack for the Open-Source Cloud
OpenStack is an open-source initiative to develop an open cloud infrastructure. First launched by NASA and Rackspace, OpenStack is designed to maximize interoperability and speed development. IBM has built a number of software products around OpenStack since making a formal commitment to the organization in 2013, including PowerVC*, PowerVM* NovaLink and its cloud orchestrator. Open development ensures interoperability and optimizes software quality, thereby fostering widespread adoption across a wide swath of industries.
Kristin Lewotsky is a freelance technology writer based in Amherst, N.H.More →