“Dedupe everywhere” is becoming a guiding principle in how enterprises now strategically look to deploy data deduplication into their organizations. But it is now recognized that these organizations need a common mechanism to deduplicate data throughout their environment to optimally manage and scale it. It is in this context that HP recently announced its new StoreOnce B6200 Backup System along with a number of StoreOnce feature enhancements as HP takes its deduplication story further into the enterprise.
Approximately 18 months ago HP announced that its StoreOnce technology would function at HP’s core deduplication technology across its product portfolio and, in so doing, help drive out the current levels of complexity and overhead that result when using multiple deduplication processes and products in today’s Dedupe 1.0 world.
However to date HP’s offerings that were based on its StoreOnce technology were relatively modest. HP only offered its D2D2500, D2D4100 and D2D4300 series of products that were primarily suitable for use in remote and branch offices as well as mid-sized data centers.
In the second round of StoreOnce related announcements HP takes the StoreOnce deduplication story decidedly upstream. Its release of the StoreOnce B6200 Backup System coupled with a number of significant feature enhancements to its underlying StoreOnce technology confirm HP’s commitment to both extending StoreOnce throughout its product portfolio and capitalizing on prior technology investments. In so doing HP seeks to drive Dedupe 2.0 (a singular approach to data deduplication) throughout enterprise organizations.
The StoreOnce B6200 Backup System illustrates how HP intends to deliver on this objective. The StoreOnce B6200 Backup System may be best defined as a crossroads product within HP’s product portfolio as beneath the hood it bears a striking resemblance to existing HP products and technologies. For example:
- The B6200 may consist of either standard HP Proliant servers or blade system hardware
- The B6200 merges the StoreOnce deduplication algorithm with HP’s existing IBRIX file system
- The B6200 also leverages IBRIX’s clustered architecture to facilitate the introduction of high availability, or “autonomic restart,” into the B6200’s design so that one backup node can fail over to another in under one minute
Where the B6200 begins to differentiate itself from HP’s other product lines is in its introduction of a number of new features that reflect HP’s Dedupe 2.0 vision for the enterprises. For instance, the B6200 introduces a new technology that HP refers to as “couplets.”
B6200 couplets may be deployed in a configuration as small as 48 TBs raw (~32 TBs usable) which is the next logical configuration size above its upper end D2D4300 series. Each B6200 couplet consists of two controllers and two disk shelves to which two more disk shelves may be added before another couplet is needed. By combining couplets, the B6200 can scale up to an aggregate size of 768 TBs raw (512 TBs usable) the B6200 can scale out to centrally store all of an enterprise’s deduplicated data in one physical location.
Like any deduplication solution, the deduplication ratio that each organization achieves using the B6200’s StoreOnce deduplication algorithm will vary but, assuming a conservative 15:1 deduplication ratio, organizations could conceivable store nearly 8 PBs of data in a fully configured B6200.
However that 8 PB estimate may be very conservative considering how the StoreOnce algorithm works. Its SHA-1 algorithm deduplicates data using a variable 4K block that minimizes disk I/O and memory usage even as it optimizes storage capacity. Further, it expedites restore speeds by not replacing small amounts of duplicate data with pointers to data that is not adjacent to existing deduplicated data. This combination of features results in deduplication ratios that are as high as 35:1 with restore speeds that are comparable to what might be achieved restoring data stored in its native format. (Note: These numbers are based on estimates that HP achieved in its internal testing.)
Yet one of the motivations for designing the B6200 to achieve these fast restores, high availability, high deduplication ratios and store large amounts of data is to facilitate the centralization and management of data stored in remote and branch offices and regional data centers on the B6200.
So to accomplish this, the B6200 introduces support for the fan-in and central management of data from up to 384 sites from any of its existing D2D products (D2D2500, D2D4100 or D2D4300) to the B6200. This new fan-in feature coupled with the StoreOnce technology used on all of these products helps to optimize data storage across the enterprise even as it minimizes the amount of data sent when data is replicated.
The final piece of this current phase of HP extending its StoreOnce reach further into the enterprise is HP porting and embedding its StoreOnce engine into its Data Protector backup software. This federated approach to deduplication was done so data could be deduplicated at those sites and on those applications where a dedicated hardware target is not needed. Further, because Data Protector also uses the same StoreOnce technology, HP can now extend the same end-to-end deduplication benefits to every chunk of data in the enterprise from the largest application to the smallest.
“Dedupe everywhere” may not yet be reality for any enterprise organization but the second major round of StoreOnce product announcements and enhancements from HP certainly move it further down that path. By introducing the B6200 Backup System for the enterprise and adding the StoreOnce deduplication engine into its Data Protector product, HP for the first time has the full suite of end-to-end deduplication solutions that enterprise organizations need to optimize their backup data stores.
Yet the real differentiator for HP over its competition will likely be just far it can successfully reach into its various storage systems with its StoreOnce technology. However to know the answer to that question we will have to wait for a future StoreOnce announcement to find out.