Each year, a handful of storage technologies seem poised to break out of the pack and become essential building blocks for new products that make storage easier to manage, less costly and better performing. For our annual forecast, these are the five technologies we think will be hot in 2009:
- 10Gb Ethernet (10GbE) and 6Gb/sec SAS are less-expensive alternatives to Fibre Channel (FC) networking and storage
- Remote replication for disaster recovery, while not new, is becoming the cornerstone of DR plans.
- Global deduplication, managing islands of dedupe and virtual tape library (VTL) appliances and sharing dedupe data among them, is a much-needed innovation for next-generation dedupe products
- Storage-as-a-Service (SaaS) offerings are becoming increasingly appealing in tough economic times.
- Self-healing systems, arrays that help cut management time and data loss, round out our list.
As we do each year, we'll cite several promising technologies that, for various reasons, will not be hot in 2009 (see "Not hot in 2009," below). Finally, we'll do a self-imposed reality check and rate the accuracy of the hot technology predictions we made last year (see "Report card on our 2008 predictions").
SAS-2 spec (6Gb/sec) and 10GbE
These are different technologies, but together they put pressure on FC's dominant position in storage networking. At 10Gb, Ethernet becomes more of a storage play because it boosts iSCSI's performance. And the serial-attached SCSI (SAS-2) standard will fortify that interface as "enterprise class" in 2009, largely because of its 6Gb/sec capabilities.
That means SAS should start to challenge FC for high-performance disk drives. Analysts say 6Gb/sec SAS will eventually become the new enterprise storage drive standard. "It's going to be the first time that this serial-attached SCSI interface is faster than FC," says John Rydning, research director for hard disk drives at analyst firm IDC. "We've seen a pretty rapid adoption for internal storage, and now it's going to get the attention for external storage," he says.
Doubling from its current 3Gb/sec bandwidth, 6Gb/sec SAS enables solid-state disk (SSD) adoption and compatibility with the SATA connection. In October, LSI Corp. brought out what it calls the industry's first 6Gb/sec SAS-to-SATA bridge cards and 16-port SAS storage processors.
Marty Czekalski, senior staff program manager at Seagate Technology and VP of the SCSI Trade Association board of directors, says 6Gb/sec SAS will start shipping in systems to customers about halfway through 2009.
"You're still going to have FC SANs between servers and external storage systems, but you will start to see 6Gb/sec SAS drives used on the back side of those controllers. They will replace the FC drives on the back side over time," notes Czekalski. Less cabling, doubled transfer rates, improved link utilisation and rack-to-rack distances are among the advantages 6Gb/sec SAS users can expect, he says. "And 6Gb/sec SAS is a great connection for SSDs," he adds, because users can get 6Gb/sec per link, low latency and high aggregate performance.
Brad Booth, president and chairman of the board of the Ethernet Alliance, says Ethernet has progressed from a technology that would carry only LAN traffic to "a unified fabric ... iSCSI, NAS, FCoE all rely on the Ethernet," he says. Booth, who's also senior principal engineer in the office of the CTO at AMCC, says the advent of electronic dispersion compensation (EDC), used in optical and backplane platforms as a means to compensate for some of the impairments in the transmission medium, is among the most recent developments giving 10GbE technology a boost.
The biggest obstacle for 10Gb remains price, but industry analysts agree that the price of 10Gb in 2009 will drop as the technology matures. In addition, SFP+, a new optical form factor, will permit greater port density and lower the price per port.
Data replication for DR
The use of remote data replication for DR isn't new, but it has taken off in large part to the increased role of server virtualisation. It was more than a decade ago when storage firms began offering capabilities in their storage arrays to copy or replicate data to a remote storage system, but that ability often relied on having identical server and storage hardware at both locations.
Today's server virtualisation technology, which allows the same images to run on different server types at each location, is simplifying replication for DR and lowering its cost. The "game changer" for replication in 2009, according to Stephen Foskett, director of data practice at consultancy Contoural, is the combination of widespread server virtualisation and virtual storage technology "and, most importantly, universal APIs and management systems to stitch the two together. VMware's SRM [Site Recovery Manager] is a good example of the kind of end-to-end technology that will finally allow storage replication to become a standard component of the data centre," he says.
Foskett predicts new products will make it easier for small- to medium-sized businesses (SMBs) to replicate data to DR sites, and for enterprises to move data generated from remote offices to their main data centre. In 2009, the bottom line is that replication for DR will become more affordable. According to Greg Schulz, founder and senior analyst at StorageIO Group, replication technology will no longer be for the "rich and famous," and will become more prevalent in organisations of all sizes.
Deduplication is growing up and out. Global dedupe expands upon "plain vanilla" dedupe technology by working across multiple processors and dedupe appliances, which can share data. The technology creates a single dedupe database that can be managed from one console. This is a big deal because instead of trying to manage a bunch of separate deduplication appliances, a dedupe system can scale to meet the needs of growing companies and exploding storage systems.
In 2009, there'll be more users like Eric Zuspan, senior system administrator, SAN/Unix, at MultiCare Health System. Zuspan purchased Sepaton Inc.'s S2100-ES2 Enterprise VTL with DeltaStor software for data deduplication this year in hopes of easing his reliance on tape. His shop also owns a Data Domain Inc. appliance (which is handled by the WAN team that serves all the Windows applications). What Zuspan found in Sepaton is a single appliance that has multinodes, and the ability to add more nodes if he needs extra capacity or throughput.
"This appliance gives us that kind of flexibility," says Zuspan. (Rather than using the term multinode, the Storage Networking Industry Association refers to the capability as "single deduplication domain." This means that all data delivered from any node in a system participates in the same deduplicated pool of storage.) Data Domain says it expects to support clustered nodes in 2009.
The new and improved Storage-as-a-Service (SaaS) offerings look a lot different than they used to, and are getting a lot of attention. EMC and IBM are among the storage heavyweights who have made significant SaaS acquisitions in recent years. This has 2009, with all its looming economic pressures, shaping up as the year that SaaS (the outsourcing of backup operations to a company with its own hosting facility and software to manage it) will become even more appealing. Whether the selling point is a "cloud" or an underground vault, SaaS will get plenty of attention from storage pros who may find it less expensive to engage a SaaS provider than to protect all or part of their data in-house.
SaaS is also a way to address pricey compliance requirements and new DR mandates. New data-encryption transfer technology and tiered SaaS offerings from vendors are making it an attractive alternative in shops where security concerns kept some IT executives from considering the possibility of shipping their data offsite. Is SaaS still targeted for SMBs? Yes, but large companies are starting to seriously consider SaaS because of the requirement to back up data from a growing number of remote offices and a new awareness of the importance of backing up critical information on laptops, says Stephanie Balaouras, principal analyst at Forrester Research.
There have been a lot of SaaS company acquisitions and repositioning recently, setting up 2009 as the year when users will have more SaaS choices. Dell made a $US155 million acquisition of MessageOne, and company executives say they plan to use the newly acquired SaaS technology for remote data protection and systems management. CommVault extended its managed services agreement with smaller SaaS player Incentra Solutions. In September, Seagate announced i365, A Seagate Company, a new umbrella company designed to bring together the service businesses Seagate has acquired in recent years. Companies such as AmeriVault, Intronis Technologies and Seagate (EVault) are all competing on things like laptop support, on-demand restores, open-file management and multiple data storage facility locations.
One caveat to this SaaS prediction is that some smaller SaaS vendors, already practically giving away their services to establish a clientele, are likely to be squeezed out of the SaaS picture as those veterans with the most clout (EMC, IBM, Iron Mountain Digital and Seagate) continue to mine this opportunity. "I see some of the hosting companies becoming pressured," says StorageIO Group's Schulz.
There's plenty of noise being made in this arena and it can be tough for users to decide what to focus on with so many vendors shouting so loudly. What can a self-healing system do for you? It can eliminate a potential single point of failure in a RAID array, for one. Self-healing systems reduce the risk of data loss on a disk drive caused by media defects by inspecting adjacent areas around the first defect. The system then reconstructs the data associated with the first defect using parity. All of this can be accomplished in the background to allow the host uninterrupted access to data.
"The whole point of these systems is to reduce maintenance windows and to prolong technology investments," says Brian Babineau, senior analyst at Enterprise Strategy Group.
In April, Atrato and Xiotech introduced storage systems with a new pledge that customers could expect years of operation without needing any service. Pitching their products as self-healing systems with sealed components containing multiple disk drives, both vendors say they avoid RAID rebuilds by copying data off a troubled drive and, depending on whether the drive has failed, replacing it and copying data back to the new or repaired drive.
These so-called no-maintenance disk arrays are designed to avoid hard-drive swapping. Self-healing systems have been getting their share of press for many years, but as storage budgets tighten, any array that cuts the amount of time storage professionals spend maintaining and fixing troubled disks is time (and money) saved that can be spent on other data protection issues.
Self-healing storage technology isn't a new concept and some storage pros will remember IBM's Shark, its TotalStorage Enterprise Storage Server with self-healing capabilities that was introduced in 2002; yet it wasn't until 2005 that IBM said the "era of self-healing technology" had arrived. And archiving systems like EMC's Centrea have been billed as self-healing for years.
We may be going out on a limb with this prediction, but we're betting that a demand for quicker, more efficient rebuilds and increased pressure to provide 24/7 support for business processes will propel self-healing systems into a "must-have" feature for all storage arrays in 2009.
- Fibre Channel over Ethernet (FCoE). It isn't sizzling yet. Sure, it was early 2008 when Cisco closed the deal on its Nuova acquisition and unveiled the Nexus 5000 series FCoE unified fabric switch. We also know that Cisco--along with Emulex, Intel, QLogic and others -- has cranked up the heat to declare FCoE ready to take off in 2009. But we're betting FCoE won't start picking up traction until 2010 at the earliest.
- Green storage. Once IBM started making catchy commercials parodying the gap between IT and corporate when it comes to green initiatives, it was clear the green revolution was in trouble. But IBM has it right: Green storage is about saving money. Green storage technology needs to work its way into storage products more tightly, and we need some good, easy-to-compare benchmarks for storage managers to address conservation issues. That will take time.
- InfiniBand for storage. It sure gets enough press, but it still hasn't caught on for mainstream storage. It's currently used mainly for high-performance computing on the networking side. Is it holding its own? Yes. Is it heating up? For now, it's lukewarm.
- (And Five) Primary storage dedupe/compression. NetApp and two smaller companies (Ocarina Networks and Storwize) have been touting their ability to bring dedupe and compression technology to primary storage, but it's not being used for high-transaction database data because deduplication works best with data that doesn't change much; performance is also an issue.
Deduplication: This was a slam dunk, kind of like one of those easy electives you take in college to boost your GPA. Deduplication was all the rage this year for good reason: It's a proven technology that pays an immediate dividend. The products and technology are maturing, and users are often seeing the dedupe ratios that vendors promised.
eDiscovery: Everyone is talking about it, but not everyone is buying new tools or developing new strategies to deal with it. In part, we're waiting on storage managers to get used to new roles as managers of corporate information vs. managers of the systems that host that information. Outside of law firms, storage pros are just catching up to the painful realities of litigation holds and eDiscovery requests.
Green storage: We'd like to grade on a curve here. The fact that green storage is on our "not hot" list for 2009 is pretty good evidence that we didn't exactly ace this one. Then again, most storage pros couldn't avoid hearing a green pitch from a vendor--or being handed a corporate green initiative--this year. In that regard, our prediction was dead on. Is it our fault the vendor messaging is missing the mark? We didn't anticipate the wide divide between vendors and the IT pros being asked to open their wallets (not just their minds) to green data centres.
LTO-4: Adoption is increasing according to our surveys. But so far, there hasn't been a mad rush toward LTO-4, and LTO-3 is still more prevalent. LTO-4 is almost an inevitable upgrade when companies decide to refresh their tape technology and want to take advantage of advanced features such as encryption. We were too exuberant in 2008, but we'll take credit for endorsing the technology.
N_Port ID Virtualisation (NPIV): The jury is still out. VMware and Microsoft support NPIV, which lets virtual server guests see a dedicated partition of a host bus adapter in a Fibre Channel SAN. Citrix is working with Emulex on supporting NPIV and Virtual Iron Software has certified with Emulex. Still, it's not exactly at the top of storage managers' priority lists.