As DCIG readies its third release of the DCIG Deduplicating Backup Appliance Buyer’s Guide, it always encounters certain trends and the emergence of new features in the products covered in each respective Guide. DCIG’s experience was no different in its preparations for this Guide. Virtual appliances and scale-out and scale-up architectures in particular caught our eye as DCIG prepares to release this Guide.
Deduplicating backup appliances come in two forms: physical and virtual. While the forthcoming Buyer’s Guide covers physical deduplicating backup appliances which offer hardware and software as a single SKU, virtual appliances utilized for deduplication represent a growing trend. This year, DCIG chose to examine which physical deduplicating appliances are also available as virtual appliances.
Three of the vendors in the forthcoming release of this Buyer’s Guide offer virtual appliances: Quantum in its DXi V1000 and V4000; HP with its StoreOnce solution; and Dell makes available the DR2000v. Some virtual appliances geared for deduplication may operate as a standalone solutions while others can only be deployed with a complementary physical deduplicating backup appliance. For example, Dell’s DR2000v may only be used when it is deployed in conjunction with either its DR4100 or DR6100 models. In every case where a vendor offers a virtual appliance option, it can interact with physical deduplicating backup appliances in that vendor’s family of products with the virtual appliance providing most if not all of the same deduplication and management capabilities as the physical appliance offers.
Small and midsized enterprises (SMEs) often have remote and small offices that are now highly virtualized and require an easy to deploy, easy to manage and cost effective deduplication target. This is where virtual appliances have particular appeal. When the last DCIG Buyer’s Guide on deduplicating backup appliances was produced, the only deployment option that remote and small offices realisticaly had was to use smaller, physical deduplicating backup appliance (which are still available and will be covered in this Guide.)
Now, instead of purchasing a physical appliance for these offices, they can purchase a virtual appliance which are generally available at a lower cost. At time of publication, the Dell DR2000v virtual appliance listed for $4,200 which includes one terabyte of licensed storage capacity, $7,500 for two terabytes of licensed storage capacity and $13,500 for four terabytes of licensed storage capacity. The HP StoreOnce virtual appliance with four terabytes and ten terabytes of licensed, usable capacity are priced at about $2,000 and $5,000, respectively.
One benefit of the virtual appliance is it removes the need for any additional physical hardware. Virtual appliances support the leading hypervisors such as VMware ESX and Microsoft Hyper-V and may be installed on existing hypervisor servers.
Scale-up, Scale-out or Both
Two methods for adding storage capacity to a deduplicating backup appliance exist: scale-up and scale-out. Some appliances are even capable of both which give organizations the flexibility to scale up by adding internal storage and then adding more nodes to scale-out which remaining in a single logical configuration for simplified administration.
Deduplicating backup appliances that only offer a scale-up architecture have a wide range of usable capacity going from single or low double-digit terabytes of capacity to more than 100 terabytes. This compares to deduplicating backup appliances that only offer a scale-out architecture where nodes with preconfigured amounts of storage capacity are added to the existing configuration.
Each architecture has its benefits and limitations. Using a scale-out architecture, organizations purchase nodes as they need them. Each time they add a node to the solution, it provides more storage capacity, network interfaces and processing power. In a scale-up configuration, each component (processing, networking ports and capacity) are scaled incrementally. As such, the number of processors does not necessarily increase as more storage capacity is added.
One benefit of using a scale-out architectures is that the nodes are viewed and treated as one single logical entity by the solution. However just because they are all managed as a single, logical solution does not necessarily mean all of the nodes work together as one. Organizations need to verify that a deduplicating backup appliance that uses a scale-out architecture also offers “global deduplication” which deduplicates data across all of the nodes in a multi-node system. If it does not offer this feature, data is still deduplicated but only on each node so data deduplication is not optimized.
Another potential drawback to using a scale out architecture is the possibility of “node sprawl” as adding nodes is easy to do but it may not be the most optimal way to grow. To counter this, organizations may want to purchase individual nodes with more capacity and processing power.
ExaGrid and NEC are two notable providers that offer a scale-out architecture as part of their solution. ExaGrid recently increased the number of appliances available in a grid format from ten to 14 with its 4.7 software release, which increased its limits across its product line.
Instead of ten ExaGrid EX21000E appliances in a single grid for 480 terabyte raw capacity, it can combine 14 ExaGrid EX21000E appliances for 672 terabytes of raw capacity. NEC, which sells its products in blocks under set SKUs depending on the market segment, uses hybrid and storage nodes in various configurations to scale from as little as 12 terabytes to 7.9 petabytes.
With the consolidation of vendor lineups, a similar change is happening to models with scale-up architecture. For example, instead of selling several appliances into the mid-market datacenter market, Quantum now just sells one model for each — the DXi 4700 for the midrange and the DXi6900 for enterprise — by providing a wide range of scale-up options. The DXi 4700 scales from five terabytes to 135 terabytes and the 6900 ranges from 17 terabytes to 510 terabytes of usable backup storage.