Introduction
NetApp continues to innovate in unified storage and data management with the release of ONTAP 9.16.1. This latest update brings new features to enhance efficiency, performance and security across business-critical enterprise workloads. The newly announced back in September the ASA platforms -the ASA A70, A90, and A1K - particularly benefit from expanded capabilities. VMware administrators will see some significant quality of life benefits and their workloads attain higher levels of efficiency with this latest release. And high-performance workloads such as EDA, video workflows and AI will see performance improvements from the new capabilities in 9.16.1 with more efficient handling of large file distribution. So lets get into all that is new and glorious with ONTAP 9.16.1.
Advanced data distribution
Most of you are probably familiar with our FlexGroup volume feature. This allows for member volumes from nodes across a cluster to be brought together into a single name space volume that can reach absolutely massive scale (60PB and Billions of files). When files are written to a FlexGroup volume, they are distributed across all member volumes on a file by file basis. This also results in load balancing as groups of files are accessed from multiple nodes rather than just one node. But there are a group of workloads that create, store and access very large files such as EDA, video workflows and AI training. In these workloads files can reach the multi-terabyte size. Striping files in this case does not lead to efficiency. Our new advanced data distribution capabilities address this by breaking very large files (greater than 10 TB) into chunks and distributing those chunks across the member volumes of a FlexGroup volume. This results in reduced bottle necks and more predictable performance for those workloads. You can read more about this new capability in the technical deep dive blog here: FlexGroups and Advanced Data Distribution
SnapMirror active sync provides bi-directional synchronous replication and transparent application failover. With symmetric active-active architecture we are able to offer customers simultaneous read and write access for high availability and load balancing while also supporting both uniform and non-uniform host access.
In ONTAP 9.16.1, we are improving SnapMirror active sync in a few ways. First, we will offer a simplified experience with System Manager and ONTAP Tools for VMware vSphere (OTV). System Manager comes with enhanced and simplified workflows such as automated host access configuration, along with this OTV provides easy life cycle management for VMware admins with a native vSphere plug-in. Additionally, SnapMirror active sync supports clustering applications such as VMware Metro Storage Cluster and Windows Failover Cluster providing robust protection for your workloads.
These enhancements will help customers achieve higher scale and better performance allowing them to distribute their workloads across a larger cluster without compromising on the ability to protect their mission critical workloads against disasters.
Next, for unified AFF and ASA platforms, starting with ONTAP 9.16.1, SnapMirror active sync will support active/active configurations across up to 4 nodes. This will offer increased performance and efficiency while continuing to provide read and write access across both sites.
Block and database enhancements
With the release of the new ASA platforms at Insight in September we are seeing more and more customers turn to NetApp for dedicated block workloads such as database and VMware. We continue to innovate in our block workload support. For the new ASA platform, we are expanding support and key features in several ways.
- Active/Active pathing for NVMe – Symmetric active/active pathing for NVMe, similar to the active/active pathing supported across fiberchannel and iSCSI. Same capability but now for NVMe. This enables SnapMirror active synch but also more for fast fail over for cases where IO is interrupted going down a path of one of the nodes.
- Storage pool capacity and performance rebalancing (ASA node space rebalancing) – Over time, nodes in an HA pair that are assigned as primary for LUNs may end up being over assigned. For example, one node in an HA pair may be assigned as primary for all of the most high-performance LUNs creating a potential for bottle neck. This rebalancing resets the node preference and rebalances the LUNs across both nodes in the HA for preference. Note, there is no data movement here as the data is shared by the nodes in the HA pair. Users can opt in for self-optimization or they can opt out and run manually on alert.
- Scale out to 12 nodes or 6 HA pairs for ASA A1K, A90, and A70 – Scaling from a single HA pair to multiple HA pairs will be available with this version of ONTAP
- Increased LUN support to 10K LUNs for ASA A1K, A90, and A70 – This is an increase in the number of supported LUNs from 2K at launch to 10K LUNs per HA pair. A full 12 node cluster will support 30K LUNs.
Other ONTAP enhancements – lightning round
But wait, there’s more! In addition to all the goodness we’ve shared so far, our unified customers will see several new features and benefits in ONTAP 9.16.1 that will help various workloads including VMware and beyond. Lets do a quick lightning round on those.
- Multi-part upload for ONTAP S3 for file object duality – Multi-part upload allows “breaking” of large objects into multiple parts as part of uploading to an S3 bucket. It Significantly accelerates large object uploads and is required for compatibility of most S3 applications. We are now adding support for NAS volumes sharing their data via S3 protocol to do multi-part upload. A growing number of applications using S3 require this API for basic functionality.
- Latency and throughput metrics for volume qtrees – This feature allows for latency and throughput metrics to be shown per volume (rather than at the aggregate level). Customers are no longer required to make a tradeoff between that granular reporting and using the FlexGroup for management.
- NVMe storage deallocation (for VMware) – Also known as “hole punching” this feature allows VM admins (through vsphere) to release capacity ear marked for a VM back to the pool of available storage when the VM is deleted freeing up capacity for other uses.
- TLS (Transport Layer Security) in-flight encryption for NVMe – This secures/encrypts traffic that is transferred between the storage system and the host (VMware host). While not a major attack surface, it is a potential risk and securing the traffic between storage controller and hosts helps to harden the system. We have offered this over other protocols already, this is an expansion for NVMe over TCP. NVMe over TCP is going over ethernet switches and the traffic could be exposed so this hardening is important specifically for this use case.
- Improved upgrade experience – Reduced reboots with the newest platforms going from as many as 3 to 1 during the software upgrade process.
Conclusion
As you can see, ONTAP 9.16.1 delivers and extensive set of features and enhancements to customers across a broad set of use cases for both structured and unstructured data. If you have questions about anything you’ve read here and would like to learn more, reach out to your partner or sales rep or join us at NetApp ONTAP.
