Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
Also see the first installment in our Hybrid Cloud Storage Buying Guide.
Cloud computing is a maturing technology but cloud storage is lagging behind – with hybrid cloud storage as the most forward looking technology. The lag in cloud storage reflects the same situation between on-premise servers, networks and storage: servers and networks have outstripped storage performance many times over. There are a number of reasons for this, many of them based on the physical limitations of magnetic disk. Flash and optimizing data are helping to close the gap. Yet there is still a gulf between on-premise and cloud-based storage due to latency and low bandwidth.
Sending backup and archive to the cloud relieves some of the pressure of storing massive data. However, on-premise production data can rarely take advantage of the cloud and its scalability.
Hybrid Cloud Storage to the Rescue?
Hybrid cloud storage is a way to extend on-premise production storage to the cloud. Note that is “extend” and not “replace.” Cloud-based storage cannot yet replace high-performance local storage systems, but it can act as a semi-active tier where application servers can restore data from the cloud without using backup catalogs or archival restore procedures.https://o1.qnsr.com/log/p.gif?;n=203;c=204650394;s=9477;x=7936;f=201801171506010;u=j;z=TIMESTAMP;a=20392931;e=i
This architecture enables IT to extend the storage infrastructure to take advantage of the cloud’s scalability for application data without sacrificing performance or central management. The cloud expands on-premise storage with integrated data tiering and automated storage management across on-premise and cloud storage tiers. The integrated system also supports remote DR with backup and archiving on a long-term retention tier.
Hybrid systems generally consist of a high performance storage system for on-premise applications, ideally flash or hybrid flash storage. Local cache, accelerated WANs and/or dedicated bandwidth enable the system to quickly connect to the cloud as a third storage tier in addition to backup and archiving traffic. Vendors implement their hybrid solutions in different ways: virtual appliances, proprietary storage systems, and hybrid storage controllers are all in the market today.
What Hybrid Cloud Storage Can Do
The hybrid model is in its beginning stages. It cannot yet adequately support intensive applications with their low latency and high I/O processing requirements. However, the biggest data producers out there are not Tier 1 applications but Tier 2: SharePoint, MS Office, Exchange, and server virtualization.
We’re talking a lot of Tier 2 data: it’s growing 60% or more every year over year, faster than any other type of corporate data. This speed of growth costs a lot in capital and maintenance expenses.
The problem is that fast-growing data slows down production servers and nearline storage tiers. Even as the data ages, much of it needs to remain available to the application servers. IT may backup or archive older data to the cloud. But restoring data first before the application can access it creates unacceptable lag.
Enter hybrid cloud storage. It doesn’t try to be a primary tier – development isn’t there yet – but acts as a storage tier for less-active data. The data remains immediately available to the on-premise application, which treats the cloud storage as merely another storage tier. Both on-premise and cloud storage tiers are under IT control, which uses the same set of storage management tools across the whole infrastructure.
What to Look For in a Hybrid Cloud Storage Product
The few vendors who are really doing hybrid cloud storage vary in their implementations. Virtual appliances, physical appliances, proprietary storage systems, and virtual storage controllers all enable hybrid cloud storage. Here are the features that you should look for when considering any of them:
Low latency: Particularly with block data traffic, low latency is critical to make the cloud work as an active storage tier. Fast ingestion and restores are critical for success. Hybrid cloud storage vendors work to lower latency on both on-premise and cloud sides. The on-premise side uses efficient Web protocols, minimizes data with dedupe and compression, and optimizes the WAN. On the cloud side, some vendors present multiple cloud instances as a single resource to the on-premise application, which speeds up data performance. Other vendors map metadata across the infrastructure to accelerate data movement between the on-premise storage system and the cloud tier.
Centralized management: Centralized management lets IT manage both the on-premise storage system and its cloud tiers as a single infrastructure. These tools are simultaneously aware of both environments, and the more automated they are, the better for IT. The tools operate on both sides of the fence: on-premise to manage storage and tiering, and on the cloud to manage storage, metadata, application requests, and cloud instances.
Scalability: Hybrid cloud storage enables IT to use the cloud’s extreme scalability for production data as well as backup volumes. This level of scalability lowers storage costs because it can scale production data across the cloud, instead of IT investing in expensive on-premise storage.
Vendors in the Hybrid Cloud Storage Space
Most backup vendors already provide the cloud as a backup and archiving target. Hybrid cloud storage should include remote DR and archiving, and add the ability to store active data for on-premise applications.
Cloud providers are also an important part of this hybrid equation. The cloud provider must support not only high scalability but also high performance, and enable shared management tools to run on their customers’ stored application data.
A leading vendor is Microsoft StorSimple, which is iSCSI block-based storage that extends the application data environment to the cloud. StorSimple works with public clouds like Amazon and is optimized for MS Windows Azure. TwinStrata CloudArray offers physical or virtual appliances for storing primary and nearline data, and storing tertiary application data to the cloud.
Nasuni storage controllers and virtual appliance enable hybrid cloud storage for both block and file data. Nasuni replicates snapshots to the cloud. Panzura’s storage controller is available as a hardware appliance or virtual machine, and is optimized for CIFS and NFS workloads. CTERA’s appliances support both NAS and iSCSI protocols for file- and block-based storage.
This small vendor list once included Nirvanix, who declared bankruptcy in 2013. Developing hybrid cloud storage needs massive financial and development resources given heavy technology demands, active partnerships with the cloud providers, and intensive marketing. Not every vendor is going to make it. However, hybrid cloud storage has strong benefits for business and we will see more and more vendors take the plunge. The ones that survive will thrive in a fast-growing cloud storage market.
Photo courtesy of Shutterstock.