Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
These integrated on-premise and cloud technologies offer acceptable performance levels for Tier 2 data that can be separated by activity and access frequency. They are not yet developed enough for fast I/O requirements on the Tier 1 application side, nor are they optimized for newly created data from Tier 2 applications. But they do add immense and cost-effective scalability to the production environment and its aging application data.
You won't find HCS systems on every street corner but they are fast gaining market traction. However, buyer beware of cobbled-together systems that do not integrate on-premises with cloud-resident data.
Real HCS systems are optimized to deliver high performance for working set data and treat the cloud as an additional tier for less-active data, not just a backup target. Look for the following capabilities in any HCS system:
• Centralized management tools. HCS enables centralized management for the same storage system whether tiers are on-premise or in the cloud. Management services run in both environments to monitor and share information to the central console. The cloud becomes a storage tier instead of a separate backup or archival target.
• Efficient disaster recovery. HCS systems establish efficient links with the cloud tier to enable fast data movement between the cloud and on-premise system. The cloud tier is optimized for disaster recovery, enabling customers to quickly restore data from the cloud without the expense of remote DR sites or massive tape vaults. (The HCS system should be geographically unbound so that the restore is not limited to a specific cloud data center.) Instead, data transparently moves from Tier 3 onto on-premise storage, without requiring additional management or restore steps.
In addition, HCS systems may be able to provide application-driven restores that prioritize the data that is downloaded so high-priority applications get up and running as quickly as possible. DR testing also benefits because the amount of data involved with application-driven restores is far less than traditional tape-driven restore processes.
• High scalability. HCS hugely expands on-premise scalability by allowing production environments to use cloud's scalability. The on-premise storage system should be high performance and scalable as well, preferably offering SSD and hard disk storage with dynamic tiering. This architecture reserves the on-premise system cache, SSD and disk tiers for frequently accessed data and the cloud tier for less active data that is immediately available to the application.
There are a number of storage vendors that use the cloud as a backup target. This is a fine thing and we encourage backup admins to take advantages of cloud scalability and availability for that purpose. Archives have a similar value proposition. But HCS creates a whole new level of opportunity by enabling cloud storage for production environments.
Only a few storage vendors and their cloud partners have offerings that classify as HCS. How successful these vendors are over time will depend on the use cases they support, including backup, archiving, disaster recovery, primary storage, or all these functions.
There is a lot of confusion around backup. HCS systems are intended to back themselves up to the cloud, as opposed to being used as a backup target for a backup server or software. While target usage has value, the data in the cloud is packaged inside a backup format and is not directly accessible by applications on-premises, which means the restore process is likely to be much longer.
DR is the next of kin to backup. HCS should provide location-independent DR so that recovery operations can be done from any other location. Application data is stored in the cloud where it is immediately available for fast restore. Both Microsoft StorSimple iSCSI systems (block) and TwinStrata CloudArray (file) are excellent in this respect, and Windows Azure and AWS EC2 both support the usage case.
Using cloud storage for primary data is the most advanced use case because it requires a high performance on-premise array that manages the latency between the on-premise and cloud domains. This is where HCS has the potential to have the deepest impact on storage costs by scaling storage across to the cloud instead of bulking it up on-premises. Nasuni has a good system for file-based data, and StorSimple with Windows Azure enables primary storage as well as backup, archiving and DR with its iSCSI and block storage for the enterprise. We expect to see numerous other entries as HCS gains more and more interest in the marketplace.
The cloud partner makes a big difference as well in the HCS solution. Amazon may be good for smaller business and provides its own software gateway to provide a way to move data between on-premise customers and AWS public cloud storage.
However, enterprise customers will receive the greatest benefits from HCS even though enterprises have the most challenging environments. We find that StorSimple with Windows Azure is the best fit for the enterprise storage infrastructure. StorSimple provides a hybrid flash array with SAS disk for high performance on-premise processing, and extends storage tiering onto Azure for a highly scalable and economical storage expansion. Azure in turn provides highly scalable cloud infrastructure with cloud-based replication, geographically neutral thin restores, and data integrity scrubbing.
Taneja Group Opinion
One of HCS’ most compelling propositions is that it is a natural intersection of cloud and on-premises computing, not a technology bleeding edge. With the ability for IT to leverage cloud scalability for production environments, we expect to see sizable growth in HCS offerings. It will be interesting to see how vendors with very large scale-out on-premise storage systems react, since HCS scales storage at a fraction of their cost.
We believe that development will accelerate in response to a market hungry for cloud scalability in production environments. Ongoing development will require continual attention to security and performance concerns, but customers do not have to wait for further development to invest in HCS. HCS is here today and offers strong benefits for disaster recovery and Tier 2 primary application environments.
Photo courtesy of Shutterstock.