- Simplicity and unmatched versatility
- Comprehensive data protection
- Leading data efficiencies, such as inline compression, compaction and deduplication
- Delivers unmatched versatility that offers high performance and consistent low latency, accelerating critical workloads with industry-leading end-to-end NVMe capabilities.
- Unifies data management across storage environments — flash, disk, and cloud.
- Automates tiering of petabyte-scale datasets to object stores in the cloud or on the premises.
- Reduces deployment costs for multisite business continuity solutions.
- Accelerates read performance with cached data sets within a cluster and at remote sites.
- Increases storage efficiency savings by up to 15% for relational database applications.
- Radically increases performance by auto-tiering data between AFF systems and NetApp MAX Data on servers.
So let’s delve into some of the highlights of ONTAP 9.5.
Those of you familiar with NetApp 7-Mode will know FlexCache as a cache appliance that sat in front of ONTAP clusters. FlexCache in ONTAP 9.5 is a native capability within clusters that are sparsely-populated cache volumes, which upon reads, only ‘hot’ blocks are cached, this goes for both user data and metadata.
- Intra-cluster – caching volumes within a cluster
- Inter-cluster – caching volumes across multiple clusters
The benefits of FlexCache are:
- Higher single volume and file read throughput through multiple mounts, particularly for the single cluster deployment
- Lower read latency – local reads vs WAN access at remote sites
- Increased collaboration and productivity across multiple locations with greater efficiency, resulting in lower storage cost and simpler workflow
Some typical workload/application use cases that would make use of FlexCache are ASIC electronic design automation, media and computer generated imagery (CGI) rendering.
In this initial release:
- FlexCache is only supported in NFSv3.
- Licenses are not included in the premium bundle and are a per-cluster add-on (based on total cluster cache capacity). Subscription terms are 1 or 3 years.
- Both origin and FlexCache volumes can be provisioned on AFF, FAS, or ONTAP Select, but not all combinations make sense, so think carefully about where to deploy them.
- Both Intra-cluster origin and FlexCache volumes can be in the same SVM or in different SVMs, giving you some flexibility in deployment.
- You can only provision 10 FlexCache volumes per origin FlexVol volume, and up to 10 FlexCache volumes per node.
- FlexCache volumes are FlexGroup volumes by default, and origin volumes must be FlexVol volumes (FlexGroup volumes are not supported, neither are Qtrees and quotas on origin volumes).
NVMe Over Fibre Channel (NVMe/FC) Update
NVMe/FC was introduced in ONTAP 9.4 but that came with no storage path fail-over, which meant that customers who wanted to use NVMe/FC in production had to rely on application-based HA only if they were using Oracle, RAC, MongDB or Splunk. Support was also limited to SUSE Enterprise Linux 12 SP3.
ONTAP 9.5 introduces multi-path (storage path) fail-over with Asynchronous Namespace Access (ANA). ANA is a capability that NetApp has driven the development of, and getting it accepted as an industry standard. Similar to ALUA for FCP, ANA provides storage fail-over capabilities for SUSE Enterprise Linux 15 in this initial release; there is also support for RedHat Enterprise Linux 7.6 but without ANA.
See the NetApp IMT for HBA, switches and host software support.
Improving on the capabilities of ONTAP 9.4, which introduced quotas for FlexGroup, ONTAP 9.5 introduces capacity support for those quotas, so users can no longer over-run their assigned capacity, as well as preventing qtrees from utilizing FlexGroup vol capacity.
Both hard and soft limits can be set, and alarms can be sent to monitoring tools or via email to an administrator when limits are reached. Furthermore administrators can now obtain Qtree-level usage statistics with the statistics show –object qtree –raw command.
Further FlexGroup enhancements include:
- SnapMirror data protection best practices extended for larger deployments, so FlexGroups can now be provisioned up to 20 PiB, and the number of member volumes has been increased.
- Support for additional SMB features, including FPolicy and native file auditing, Storage Level Access Guard (SLA), copy offload (ODX), and inherited watches/changes notifications.
- FlexGroup volumes can now automatically tier to the cloud and span multiple nodes and FabricPool aggregates. Each FabricPool aggregate is linked to a separate object store and supports up to 50 billion files per HA pair and volumes as large as 20PiB. So customers can now auto-tier with FlexGroup volumes that are over 100 TiB.
ONTAP Select has also been enhanced in ONTAP 9.5 to provide dramatic gains in ESX performance in both software and hardware RAID configurations; compared with ONTAP 9.4.
Currently standard and premium licenses are available for ONTAP Select, but they are capacity tier based, where licenses are linked to a node and perpetual, and if you over-provisioned your capacity you’d be given a 30 day grace period. In ONTAP 9.5 the licenses are time-limited subscription based and are linked to a deploy Instance, where capacity limits and expiration dates are enforced, making this a very flexible deployment model. What’s great about this is that customers will no longer need to know upfront how many nodes they’ll need to deploy, giving them flexibility, simplifying licensing through a single license for multi-node deployments, and potentially reducing costs (pay as you go).
NetApp Data Availability Services (NDAS)
NDAS is a simple cloud based orchestration engine for data protection using powerful Google-like search and restore capabilities. NDAS simplifies data protection and extends data availability across the hybrid cloud by keeping active data copies in the cloud, whilst preserving snapshots and file metadata.
Only FlexVol volumes will be supported for backup in the initial release. Files and LUNS can be restored to the original source or an alternate location, whilst volumes can be restored to an alternate location.
SnapMirror Synchronous (SM-S)
Finally SnapMirror Synchronous comes to ONTAP. This is a feature that customers have been asking for for quite a while to help them meet regulatory and industry mandatory requirements.
SM-S is volume-based synchronous data replication that provides zero data loss recovery (RPO=0) and fast recovery time (low RTO).
A license must be purchased for each node in the primary cluster in addition to the standard SnapMirror license, and is priced on a per TB basis on volumes that are synchronously replicated.
Supported protocols in the initial release are FC, iSCSI and NFSv3, and other protocols will be targeted in future releases.
Round-trip network latency must not exceed 10ms and application tolerances for write latency will determine maximum practical distance over a WAN.
Source and destination volumes can be in a FabricPool aggregate, but backup policy, FlexGroup volumes and quotas are not currently supported.
The maximum number of synchronously-replicated volumes per node is 80 for AFF, 40 for FAS, and 20 for ONTAP Select.
Synchronous replication can be deployed in two modes, Strict mode, which is typically used to replicate enterprise application transaction logs, and Normal mode (default) typically used for other data volumes. The major difference between the two modes is that in Strict mode if replication to the secondary volume can’t be completed, writes to the primary volume stop, thus ensuring both volumes are always in sync and there is no data loss when failing over to the secondary site.
Typical uses for SM-S are between two datacenters, where if the primary site fails, applications can fail-over to the secondary site volumes. An example of this would be a financial services deployment over two or three datacenters, where both normal synchronous and strict synchronous modes could be used to provide complete recovery of data.
MetroCluter over IP support has available for the A700 and FAS9000 since ONTAP 9.3 and supported 4-node configurations using the Nexus 3132Q switches.
ONTAP 9.5 extends the support of MetroCluster IP to the AFF A300 (with ADP) and FAS8200 (without ADP) mid-range systems, again supporting up to 4-nodes.
SnapLock Compliance Update
SnapLock Compliance also receives an update with support for SnapMirror XDP logical replication with storage efficiency. This is a significant enhancements to SnapLock Compliance, as previously only DP SnapMirror was supported, which meant that re-synchronization after fail-over was a complex and costly operation. The benefit to customers means they will be able to perform regular DR testing with SnapLock volumes, and will be a much simpler recovery after a fail-over.
End-to End encryption is now possible using NetApp Volumes Encryption (NVE) across both performance and Cloud tiers, with a single key per volume, regardless of where the data is located. This now gives customers full control of the encryption keys and removes the requirement for server-side encryption on the Cloud tier.
- Two new object store options for IBM Cloud Object Storage (ICOS) and Amazon Commercial Cloud Services (C2S).
- Adjustable policy for adjusting how full the performance tier is before tiering to the cloud.
- Logical space used reporting to allow cloud providers to charge clients based on the amount of data stored.
- Adaptive compression is now applied to less compressible data as compared with ONTAP 9.4, enabling customers to potentially realize greater savings especially for databases.