Systems software stack gets a facelift with Power10


September 20, 2021

Timothy prickett morgan

When there is new server hardware, the system software stack should be changed to support and better take advantage of it. And now is also a great time to add new features to the system software that can be supported on an older iron. (But not too old or customers will never move forward…)

And it is the same with the various base system software – PowerVM, PowerVC, PowerSC and HMC – which run on Power Systems machines and alongside or under the IBM i operating system and technology updates. IBM i which were concurrent with the launch of the “Denali” Power E1080 machine on September 8th. The Denali machine is arguably the most powerful server in the world, or rather will be when its full 16-node configuration becomes available later this year, and this is remarkable because it is the first Power10-based machine that Big Blue. puts on the ground. (Our initial coverage of the Power E1080 can be found here, and our initial coverage of the September 2021 technology updates for IBM i 7.3 and 7.4 can be found here.)

In announcement letter 221-297, we see that both PowerVM and PowerVC have been updated. PowerVM 3.1.3 supports the Power10 iron – and presumably support for entry and midrange machines coming in 2022 is already in the code in a way so early testers can try it out. – and is also available on Power8 and Power9 systems. In addition to Power10 support, the PowerVM update can support virtual optical devices and remain bound to them during a live migration of Live Partition Mobility (LPM) from a PowerVM partition of a physical Power Systems server to a physical one. The PowerVM hypverisor now also knows how to determine which network adapter in paired systems to use as part of a live migration from Live Partition Mobility. This is important because many IBM i systems have multiple physical or virtual network ports, and you always want to choose the ones with the highest bandwidth and lowest latency for live migration.

The Virtual I / O Server, or VIOS, added to PowerVM, which as the name suggests provides a hardware abstraction layer for I / O virtualization (based on a simplified AIX kernel), also has been upgraded to version 3.1. 3 versions. VIOS 3.1.3 is also supported on Power8, Power9 and now Power10 Iron, and now includes performance improvements for virtual network interface cards (vNICs) on Linux partitions and also adds support for virtualization hybrid network for Linux partitions.

The next step for Power Stack is PowerVC, which is an IBM redesign of the Open Stack cloud controller, and with the PowerVC 2.0.2 release, IBM is moving to the “Wallaby” version that the OpenInfra Foundation (formerly the OpenStack Foundation) published in April 2021. There is often a delay of about six months between when IBM or Red Hat releases their commercial-grade OpenStack and when the open source community releases a new version. By the way, the Wallaby rollup is the 23rd version of OpenStack since the project started in 2010. The updates have been more or less on a six-month cadence, like clockwork. At this point, OpenStack is updated for fit and finish as well as for new integrations and very occasionally for new features. But it’s pretty stable software, despite having less code and more changes than a stack like IBM i itself.

In addition to grabbing the Wallaby code, IBM added a bunch of stuff to the stack, including HyperSwap support for its Spectrum Virtualize (formerly known as SAN Virtualization Controller) and IBM Cloud Object Storage (which is the result of an IBM acquisition that we cannot for a lifetime remember the name of). Other features include:

  • Thin provisioning with thresholds to help manage the physical capacity of a pool
  • Stand-alone backup node for import and export
  • Viewing additional storage and pool properties for over-provisioning and threshold
  • Using OpenStack Placement for Planning
  • Unused private flavors are periodically deleted

PowerVM, VIOS, and PowerVC software, all of which require PowerVM Linux Edition, PowerVM Standard Edition, or PowerVM Enterprise Edition, were available on September 10.

In announcement letter 221-269, the PowerSC Safety Stack for the Power Systems iron has also been updated to version 2.0, which IBM describes as “robust and simplified”. The big addition this time is the Endpoint Discovery and Response, or EDR, functionality that IBM describes as follows:

  • Host-based intrusion detection system: Analyzes traffic to and from a specific computer for signs of possible intrusion incidents, breaches or imminent threats. HIDS can also monitor key system files and any attempt to overwrite those files, a feature that works in conjunction with File Integrity Monitoring (FIM). HIDS can perform log-based intrusion detection, which overlaps with log inspection functionality, and includes time-based alerts and active response.
  • Log inspection and analysis: PowerSC 2.0 can identify important security events buried in operating system and application logs and ignore relatively unimportant events.
  • Context of the event and filtering: Events are sorted, prioritized, filtered, categorized and put into context to help identify abnormal activity.
  • Incident response: This methodology, used to respond to and manage a cyberattack, aims to reduce damage and help recover as quickly as possible.
  • Response triggers: Provides criteria for an alert that users can act on quickly.

PowerSC 2.0 can run on any Power8, Power9, or Power10 machine and requires IBM i 7.3 or 7.4, AIX 7.1 or 7.2, Red Hat Enterprise Linux Server 7.4, or SUSE Linux Enterprise Server 12 SP3. This PowerSC update was also shipped on September 10.

Finally, in announcement letter 221-217, the Hardware Management Console virtual appliance, which is the virtualized instantiation of the HMC that has been around since 2004 and controls Power Iron, has been updated to version 10.1. With the Virtual HMC 10.1 release, Power10 machines are supported, both as a stand-alone system that must be monitored as well as in core and memory pooling clusters of 2.0 enterprise pools that may contain Power8 and Power9 irons. The integration between the virtual HCM and the VIOS has also been improved and, frankly, it becomes difficult to draw lines between IBM i, the HMC, and the VIOS over time.

Virtual HMC 10.1 requires four CPUs (we assume IBM means four vCPUs, which means threads, not four power cores with up to 32 threads in the machines), 16 GB of memory, between two and four network interfaces, and at least 500 GB of space to run. HMC 10.1 Virtual Console became available on September 17.

RELATED STORIES

IBM drops Power10 in the big and bad iron first

Surprise! It’s time to refresh IBM i technology

Some practical advice on this HMC-Power9 impedance incompatibility

IBM moves HMC management to Native Power Server, LPAR

Big Blue Deploys New HMC for Freakage Power Control

The SDMC is dead, long live the HMC

IBM Terminates PowerVM V2, Encourages Customers to Adopt PowerVM V3

Kubernetes Container Control comes to power systems

IBM Brings In-House Cloud Management With New Console

IBM gives AIX some of the IBM i integration spice

IBM updates PowerVM hypervisor, PowerVC OpenStack

PowerVC Virtualization Manager Adds Promised IBM i Partitions

IBM Inflates New Cloud Controllers For Power Platforms

IBM chooses OpenStack to control future SmartClouds

Big Blue PowerVC, SmartCloud ingress provisioning for IBM i

PowerVM and IBM i Improvements Mean Better Power Systems Clouds

PowerVM hypervisor gets active memory share

PowerVM: hypervisor i is no longer hidden

About Jon Moses

Check Also

Intel promises “substantial contributions” to the growth of RISC-V • The Register

Analysis Here’s something that would have seemed odd just a few years ago: to help …

Leave a Reply

Your email address will not be published.