Tackling the Logjam in Enterprise Storage
November 10th, 2010 By: Industry Perspectives
Dan Leary is Vice President of Marketing, Nimble Storage, and has 20 years of experience as a product management and marketing executive.DAN LEARY
In the past decade, storage and backup systems have undergone many improvements. Still, there is one recurring theme across enterprises – primary storage and backup are separate processes involving separate appliances and management procedures, and they require different IT skill sets to administer them. These functions can also diverge significantly in their requirements: for example, IO performance, availability, and cost per gigabyte of storage capacity.
The logjam in storage is a result of the siloing of primary storage and backup, which, if not inevitable, is very real today. Each process, through iterative improvements, has evolved to uniquely solve the problems it faces. Primary storage delivers strong performance for the applications it has to support, and provides high uptime for continuous availability. Disk-based backup provides high efficiency and high-capacity optimization. Each process requires different expertise, and each enforces on users its own management regimen and best practices.
Siloing any two related processes is tolerable, but only until it reaches a new threshold of pain – for example, in the burden of data transfers between the two siloes, labor required, higher costs, or increased complexity. Only then do innovators step up to deal with the pain.
Silos Can Cause Issues of Cost, Efficiency
“The primary model for enterprise storage – with separate silos for primary storage, backup, and disaster recovery – hasn’t changed in decades,” said Robin Harris, founder and senior analyst, StorageMojo. “Stovepiped processes are costly to manage and inefficient, and, worst of all, they introduce latencies that put the data we want to protect at risk. But a new model of integrated and continuous data protection services is emerging. Combining storage, backup, and disaster recovery in a single process can both improve performance and reduce cost and complexity.”
Today, we’ve reached an inflection point with primary and backup storage. Primary storage is fast, but often prohibitively expensive. And, with enterprise storage requirements growing rapidly, primary storage has risen as a percentage of the overall IT budget. Disk-based storage for backup offers significant improvements over tape, but backup windows remain a challenge. We hear too many reports of backups initiated on Friday night, proceeding through the weekend, and dragging through to Monday morning. Of course, the IT staff overhead required to support these processes exacerbates the inefficiency.
Faced with a logjam in storage, innovators are now breaking it apart. One approach has been to combine primary and secondary storage in a new architecture and in a single appliance. Nimble Storage calls this “converged storage.” We accomplish this with a technology called Cache Accelerated Sequential Layout (CASL™), which makes possible high-efficiency data compression and data optimization, instant backups and restores, and WAN-efficient replication.
This new architecture has already been proven to deliver significant reductions in storage costs and management complexity by eliminating the need for separate primary and backup tiers. Technologies such as CASL can also make offsite replication for disaster recovery an integral component of enterprise storage – elevating it above an afterthought or a back-burner project. We see industry moving in the direction of converged storage and believe it’s a natural outcome of today’s less-than-efficient siloed approach.
Here are some fundamental components you should look for if you’re moving in that direction:
- Inline Data Compression. Existing primary storage architectures cannot effectively support inline compression because they store data as fixed-size blocks. Look for an architecture that supports variable-sized blocks, compressing data in real time with no added latency as these random blocks of data are written to the storage array. This approach typically reduces the amount of stored data by 50 to 75 percent.
- Optimization for Flash Memory. Flash-based devices do not have the mechanical limitations and latencies of hard drives. As a result, they can deliver very high I/O read performance, and with a modest space and power footprint. But flash storage has been prohibitively expensive for mainstream applications. Look for storage technology that allows the cost-efficient use of flash in tandem with high-capacity drives – which can cost up to 85% less than the high-RPM drives used in existing primary storage architectures.
- High-Efficiency Data Sequencing. New storage technology can deliver very high random write performance by using a new file layout designed for flash and high-capacity disks. CASL is one example; it coalesces random writes into full-strip sequential writes, and its intelligent index tracks data access patterns at a block-level granularity and updates flash within milliseconds in response to changes in application workload requirements. This ensures consistent high performance to support demanding applications such as Microsoft Exchange and SQL Server.
- Instant, Integrated Backups. New storage technology can also enable instant, application-consistent backups on the same array with up to 20x backup capacity optimization. The key is to capture compressed, incremental snapshots of the primary data on low-cost drives at pre-configured intervals. These backups are accomplished in milliseconds and have no impact on application or storage performance.
- Offsite Replication. Another key to converged storage is an architecture that can provide highly efficient off-site replication to move data to a second, remote array. This approach would mean transferring only highly compressed block-level changes, requiring low bandwidth to move data across the WAN. When remote replicas are exact copies of primary data, they can be quickly accessed in the event of a disaster by simply failing over applications to the remote array.
More simply, technologies such as CASL can eliminate the need for a separate backup tier, eliminate the complexity and administrative overhead of moving data between tiers, and make for very efficient offsite replication for disaster recovery. It merges primary and backup storage, while making disaster recovery planning a natural, easily implementable extension of an organization’s approach to data storage and protection.
When you’re evaluating your next storage purchase, consider your requirements for performance, data protection, and cost-effectiveness. Do the benefits of “desiloing” your primary and backup storage make sense for your organization?
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.