At Insight 2023, NetApp announced a new disaster recovery service delivered through BlueXP, the former Cloud Console. We take a quick look at whether this new feature makes sense for enterprises wanting to implement business continuity services.
Background
NetApp announced BlueXP Disaster Recovery as a beta preview during Insight 2023. From the limited information provided and from the on-stage presentation during the keynote, it appears that this new data protection capability operates at the storage layer, replicating ONTAP volumes using SnapMirror, presented to VMware vSphere as NFS data stores.
As a SaaS portal, BlueXP provides the benefit of automated configuration that should make the new DR solution easy to set up, whether replicating to a physical NetApp appliance or FSx for ONTAP running in AWS.
SnapMirror
The SnapMirror functionality of ONTAP is tried and tested software that reliably moves data incrementally from on-premises to the public cloud. I’ve used SnapMirror for many years for data migration and data protection purposes. When used with VMware, there are a few caveats to consider.
- Replication occurs at the datastore/volume level of granularity. Consequently, any virtual machine targeted for failover will need the entire datastore to be replicated. Depending on what’s being replicated, more data than necessary could be copied (with an associated cost).
- Failover becomes “all or nothing” for all virtual instances on the datastore or requires administrators to perform more admin work during and after a failover.
- Failover requires an active ONTAP appliance at the target site or an active FSx for ONTAP volume to receive the data. The cost of storage at the DR site has to be considered as part of the solution.
- The target of replication is another ONTAP platform; therefore, data can’t be replicated to virtual instances using other non-NetApp storage.
There are benefits to replicating at the storage layer. If the DR reason isn’t storage-related, then failback to the primary site can be achieved with the transfer of only any incremental updates while the applications were running at the DR site. DR failover is quick, as the data is in the correct format to be accessed by the hypervisor (there’s no “rehydration” from S3 images needed, for example).
Storage Agnostic
Is storage-based replication the best solution for the public cloud and on-premises? Historically, storage replication has been the most efficient and timely process for creating disaster recovery copies of data. Synchronous replication (for example) creates RPO=0 copies with no loss of data. However, the trade-off is the requirement to have like-for-like storage at both primary and DR sites. Pure Storage recently announced a DR solution but chose to use VMware data protection APIs to extract updated data from active applications and be storage agnostic.
The Architect’s View®
With the ubiquity of ONTAP on-premises and in the public cloud, NetApp offers the only viable solution for storage-based replication with widespread support for the three leading western public cloud vendors. That uniqueness needs to be contrasted with the requirement to use FSx for ONTAP in the public cloud, which is more expensive per TB of capacity than basic native block storage solutions.
As an example, FSx for ONTAP is priced at $0.145/GB-month in the EU-London region, the same as io2 storage, whereas gp3 SSD storage is only $0.0928/GB-month. While there are other cost metrics to consider, BlueXP DR is charging premium storage pricing.
We see this offering as beneficial for customers that are already embedded into the NetApp storage ecosystem. However, we also think the cost of implementation may be an issue for some customers, especially for storage that rarely (if ever) gets used for production purposes.
Then there’s the philosophical question of whether to replicate data to a virtual storage appliance in the public cloud. This decision is one we can’t answer but needs to be a strategic decision made by cloud architects. Either way, NetApp is offering choice to customers already familiar with the SnapMirror replication and failover process.
Copyright (c) 2007-2023 – Post #b47b – Brookend Ltd, first published on https://www.architecting.it/blog, do not reproduce without permission. NetApp is a Tracked Vendor by Architecting IT for data protection and storage systems. NetApp has previously been a customer of Architecting IT and Brookend Ltd.
This post is only available to Individual, Vendor or Enterprise subscribers. Restrictions on distribution are based on those licensing terms. Check our Terms of Service for details.