Hot off the software press is Nutanix AOS 5.20 LTS (Long Term Support). This is a major new release, that rolls ups new features and bug fixes from AOS STS (short term support) releases 5.16, 5.17, 5.18, and 5.19. AOS 5.20 is a maintenance branch from 5.19.
You can directly upgrade to AOS 5.20 from AOS 5.15.x, AOS 5.18.x, and AOS 5.19.x. If you are running AOS 5.16.x or 5.17.x, you will first need an intermediate upgrade to 5.18 or 5.19.
As AOS 5.20 is a ‘roll up’ of several prior STS releases, it has a big new feature payload when compared with the last LTS. However, if you are running AOS 5.19, AOS 5.20 is just a branch from that release with a longer support window and no real new features.
Listed below are the major new features and enhancements in AOS/AHV 5.20.
- Blockstore – Is an on-disk layout that is optimized for AOS. Written entirely in user space, which reduces CPU usage and overall latency. This also enables the use of SPDK for maximum NVMe/Optane performance. We are seeing approximately 20% – 25% better read IOPS and read throughput with Blockstore + SPDK. For writes, we are seeing ~10% better IOPS and throughput with Blockstore + SPDK. If you throw RDMA in the mix for writes, the performance can jump ~25% – 30%. These numbers were obtained from the 4-corners benchmark on a 4 node cluster.
- Dynamic Oplog – Oplog (which buffers random writes), can now dynamically grow beyond 6GB. This benefits nodes with fewer active vdisks. We’ve seen up to 20% improvement in DB restore operations.
- Oplog Sustained Mode – For DB workloads, up to 40% improvement in average TPM.
- AES Expansion – AES is now enabled by default on new containers on all flash cluster configs.
- Merged Snapshot Metadata – Enhances random read IOPS.
- AHV Sync Replication – Zero RPO is now supported for AHV, in addition to the already supported ESXi hypervisor.
- Cross Cluster Live Planned Failover – Ability to move running VMs from one AHV cluster to another as part of a planned failover of an entire Recovery Plan, or individual VMs at a time.
- Near-Sync Support with Metro – Metro break/re-enable does not impact NearSync relation and vice-versa.
- Near-Sync Support with Metro (ESXi) – Supports zero data loss for customers with stringent requirements.
- Near-Sync Support with Leap – Achieves feature parity between Leap and general data protection capabilities.
- Single Prism Central support for Leap – PC and PEs need to be running 5.20. Source and target Prism elements should have same hypervisors.
- Multi-site Replication for Leap – Leap DR now supports Multi-site replication. Supports private and public cloud targets.
- Network segmentation for replication – Customers can better control replication traffic.
- Identity Based Microsegmentation for EUC – Identifies desktop VMs based on user/department AD group. Allows you to define inbound/outbound policies and is secured with Flow.
- Flow Service Groups – Uses IANA ports and groups. Ability to create custom services based on your needs.
- Flow Address Groups – Ability to use user defined custom names to denote group of endpoints (IP addresses, subnets, etc.).
- Centralized Native KMS for Remote Clusters – Now supports 1 and 2 node clusters.
- Multi-Cluster key backup from Prism Central – Backup encryption keys for multiple AOS clusters with native KMS from Prism Central with one click.
- Microsoft VBS and Credential Guard for AHV – Now supported.
- Secure Boot with UEFI for AHV – Now supported
- Foundation Central – Image multiple clusters across multiple sites. Deploy hypervisor + AOS.
- New Data Resiliency Widget – Enhanced view into storage resiliency.
- Storage Over-provisioning ratio widget – New. Track oversubscription. Admins can set limit for over-provisioning ratio.
- Recycle Bin – Safety net to recover from unintentional VM deletions. Supports AHV and ESXi.
- PC Scalability – Enhanced to 25,000 VMs and 300 clusters from a single PC instance.
- CAC (Common Access Card) authentication in PC – New
- New Prism Tier: Ultimate – Adding ability to discover and monitor apps. 5..20 starts with Microsoft SQL. Adds cost metering for on-prem resources.
- Expanded Automation – 4 more trigger types, 3 more actions (add disk, expand disk, sent alert to webhook).
- AHV Multi-GPU Support – Better VDI experience
- vGPU Live Migration – New
- Improved physical network management – New logical “virtual switch” object. Replaces bridge and ‘br’ terminology.
- Cross-container vDisk Migration – New
- OVA Support – OVA navigation, export, upload, download, import, deploy.
Simplified Application Clustering – Volume groups can be directly attached to an AHV VM, removing the need for in-guest iSCSI connections. Supports Windows and Red Hat Enterprise.
As you can see from the above list, AOS and AHV 5.20 have a lot of new features when compared to the last LTS release. If you are already running AOS 5.19, then you have these features but AOS 5.20 provides longer term support. I find that many customers prefer the LTS ‘train’, so the above feature list is a big jump in performance, security, automation, and AHV capability over the prior STS release.
**Important**: There is a migration process to the new ‘virtual switch’ model. There are a few pre-reqs that customers need to understand and validate before upgrading. Check out the Virtual Switch Guide for more information (login required).