As the sheer amount of content on the Web has continued to increase exponentially, so too have the storage needs of large enterprises.
Remember when terabytes seemed big? Well with the sheer amount of content being generated today enterprises are now starting to think in terms of petabytes. And the trouble has been that all this data is not only expensive to store but it's getting more difficult to retrieve it from storage to servers on a timely basis. Thus, companies have been investing in technologies such as virtualization and cloud storage that can help them simplify their overall IT architecture, lower their costs and run their storage operations more efficiently. Deni Connor, the founding analyst at Storage Strategies Now, says that adapting to these trends is the goal of most cutting-edge storage companies on the market.
"We see a lot of companies coming out with solid-state appliances that put data into a cache and migrate that data from the cache when it's accessed less often," she says. "And everyone out there is virtualizing their servers and they're having problems getting at their data and knowing where that data is in a virtual environment."
STORAGE STARTUPS: View a slideshow of products from these startups
MORE NEWCOMERS: 7 hot cloud companies to watch
John Mascarenas, the investment director for Intel Capital, says companies that want to store data on a multi-petabyte scale will need to look toward cloud-based solutions to meet their needs since they're facing limits in what traditional physical storage can provide.
"There's not enough physical capacity for the amount of data that's being generated," he says. "It continues to move past the place where we can keep building physical storage."
The result of this, says Mascarenas, has been a large amount of innovation in a space that has not previously been thought of as a cutting-edge industry. According to analysis from Strategic Advisory Services International LLC, venture funding for storage companies totaled $458 million through the first three quarters of 2011, or roughly 42.4% more than the $321.5 million in venture funding the storage industry received in the first three quarters of 2010 and more than double the $213.7 million in venture funding storage startups received through the first three quarters of 2009. What's more, storage mergers and acquisitions have strengthened over the past two years, growing from 21 deals worth $3.3 billion in 2009 to 34 deals valued at $9.1 billion in 2010 to 23 deals adding up to $8.7 billion through the first three quarters of 2011. [For more on mergers, see our "Top tech M&A deals of 2011" slideshow]
With all this in mind, here are seven storage companies worth watching by IT organizations looking to get a better handle on their storage needs.
Founded: 2009 Focus: Storage for large unstructured data Location: Belgium
Who's the boss: CEO and co-founder Wim De Wispelaere specialized in deduplication and disk backup technologies while working as a director of product management at Symantec. COO and co-founder Wouter Van Eetvelde is also a Symantec veteran who helped build the company's PureDisk, Dedigate and DataCenter technologies.
The lowdown: When it comes to storing data, the Belgium-based Amplidata is thinking big. That's because the company's major specialization is helping enterprises store large petabyte-scale masses of unstructured data.
How big are we talking about? Consider one of the company's most recent projects, which was to help the Swiss Ecole Polytechnique Federale de Lausanne (EPFL) build a disk-based archive for more than 5,000 hours of video footage of the Montreux Jazz Festival. EPFL says that it chose the AmpliStor system because it could deliver the latency and management capabilities of disk with the cost and easy availability of LTO tapes.
"They needed a disk system that would be an archive for the university," says Tom Leyden, the director of alliances and marketing for Amplidata, which is backed by Endeavor Vision, Hummingbird Ventures and Swisscom Ventures. "The plan is long-term to make the concerts available to the public ... it will take a while but we've started to copy data onto our systems and it will take one to two years to complete."
Amplidata says that it's able to store massive amounts of data that can be quickly recalled by using its patent-pending BitSpread Codec algorithms to eliminate redundant data. Essentially, the technology breaks up the data into unrecognizable chunks and tags them so they can be reconstituted at will.
Leyden says that Amplidata's technology is essential because standard RAID storage systems can no longer keep up with large-scale data needs.
"We're seeing 3TB disks these days and since disks have become so big, the restore time has become more days and weeks than hours," he says. "This means your system becomes less protected and you have a bigger chance of losing extra data during your repair."
Amplidata is now pitching its AmpliStor system -- which is built around low-power storage nodes -- to cloud providers, SaaS application vendors and media and entertainment companies.
Founded: 2009 Focus: All-flash storage Location: Mountain View
Who's the boss: CEO Scott Dietzen was previously president and CTO of open-source messaging company Zimbra, which is now part of VMware. Founder and CTO John Colgrove was a founding engineer at Veritas Software, now a part of Symantec, where he served as CTO for the company's Data Center Management Group.
The lowdown: While hybrid cloud storage may be trendy, Pure Storage is still convinced that it can deliver an all-flash storage system that can deliver storage space at a price that's competitive with standard disk solutions. Similar to Amplidata, Pure Storage uses data-reduction algorithms that compress and deduplicate redundant data. Pure Storage CEO Dietzen says that eliminating redundant data allows Pure Storage to reduce the total amount of data stored by five to 20 times.
"Let's say that we have 38TB of virtual machines running and what actually gets stored is 2.5TB," he says. "We're storing less data than competing systems around the world."
The startup's flagship product is the Pure Storage Flash Array, which has two main parts: controllers that handle data reductions and storage shelves that have large flash storage capacities. It also features a fully virtualized operating environment that is key to limiting the amount of data that needs to be stored on flash. Among other things, the Purity Operating Environment performs global deduplication, compression and thin provisioning operations to wring out as much excess data as possible before placing it into storage. What's more, the company says that Purity "abstracts" individual flash devices into a single storage pool, thus placing data efficiently throughout the system.
Needless to say, Dietzen is quite proud of his company's achievements and thinks that its all-flash memory hardware is just what the current market is looking for.
"Most mainstream players right now are trying some kind of hybrid," he says. "We've found a way to deliver all-flash at a lower price point -- $6 per gigabit is actually our starting point."
Pure Storage announced over the summer it had garnered $30 million in Series C funding, and has now received a total of $55 million in capital investments from such outfits as Redpoint Ventures and Greylock Partners.
Founded: 2010 Focus: SSD optimization Location: Louisville, Colo.
Who's the boss: CEO and founder John Spiers is best known as the founder and CTO of LeftHand Networks, which was sold to HP in 2008. He has been working in the field of storage for 25 years and holds three patents for his innovations in storage technology. CTO and founder Kelly Long was Spiers' partner in founding LeftHand and he has also worked at a wide array of companies including Sun Microsystems, MySQL and StorageTek. He has specialized in clustered storage systems and hard drive performance, among other fields.
The lowdown: Typically, says NexGen Vice President of Marketing Chris McCall, companies face a choice when it comes to storage: They can either have more speed or more capacity. Or put another way, they can get a speedy SSD that delivers speed but creates bottlenecks or they can use a hard disk drive that will give them more space but less speed. The reason for this, he says, has to do with how SSDs are typically integrated into storage architecture.
"With a standard SSD, you get solid-state chips and you put them in a disk drive slot," he says. "But the issue is that the disk-drive connection and storage controller create bottlenecks because SSD is so fast. ... We put SSD on a PCIe bus that connects to the processor to eliminate the bottlenecks. And because we're not consuming disk-drive slots it opens up more capacity."
The company says that its PCIe-based offering can deliver "22 times more I/O" than standard SSDs that are located behind storage controllers and are 43% less expensive than SSDs behind storage controllers. And because the architecture delivers the speed of SSDs without the capacity constraints, McCall says that it can give users the ability to provision performance in much the same way that they've typically provisioned capacity. This, in turn, lets users maintain quality of service on their mission-critical data volumes "even during system events that impact performance," the company says.
NexGen CEO Spiers says that the company's architecture also removes some of the constraints that have typically made organizations reluctant to virtualize their applications and databases.
"A lot of customers have virtualized their non-critical applications and some have attempted to virtualize their Oracle database or their SQL server database and they find that the performance is all over the place," explains Spiers on a video posted on the company's website. "We allow them to provision performance for those apps in a VMware environment and make it deterministic."
NexGen announced Series A funding of $1.3 million early this year, and earlier this month said it had received $10 million in funding in a round led by Next World Capital.
Founded: 2009 Focus: Highly scalable data center building blocks Location: San Jose
Who's the boss: Founder and CEO Dheeraj Pandey previously worked as vice president of engineering at Aster Data and also managed development of storage engines at Oracle. Founder and CTO Mohit Aron was chief architect at Aster Data and also led the design and development of the Google File System. Founder and Chief Products Officer Ajeet Singh, another Aster alumnus, also worked at Oracle to help develop the company's strategy for cloud computing.
The lowdown: Nutanix's data center appliances can be used for both computing and storage. What makes the company's building blocks unique is that they're compact in comparison to other such devices and that makes them easier to stack and scale, according to the company. Each block contains four complete x86 servers that total 48 cores, 192GB of RAM and a 1.3TB Fusion-io PCle SSD.
Tiffany To, director of product marketing, says that the company wanted to combine computing and storage in appliances to drastically improve overall performance and reduce the time it takes for storage to interconnect with computing.
"Even if you add lots of IOPS [input/output operations per second] into your storage, you still have to drive those IOPS through the rack, through network switches and across to the server running your VM," she says. "What we're saying is, if you need fast interconnection, put the computing in the same building block with storage."
The result of all this, says Nutanix, isn't just improved performance but a 40% to 60% capital expense reduction due to implementing simpler, more compact architecture that the company says will reduce your overall data center footprint. And Nutanix isn't just tooting its own horn either, as the company won a gold award for desktop virtualization at this year's VMworld.
"This is what the Googles and Facebooks and Yahoos of the world do," said Partha Ramachandran, the Nutanix director of products during VMworld 2011 this year. "They have tens of thousands of servers with local disk attached, and they ... bring computing and storage together."
Nutanix in October had its approach validated further, grabbing $25 million in Series B funding led by Khosla Ventures.
Founded: 2009 Focus: Hybrid cloud storage Location: Santa Clara
Who's the boss: Co-founder and CEO Ursheet Parikh was director of product management for Cisco's WAN optimization and application acceleration business unit. Co-founder and President Guru Pangal has been working in storage for more than two decades and was one of the pioneers in developing Fibre Channel and SCSI protocols.
The lowdown: StorSimple offers a hybrid cloud storage appliance that aims to dramatically simplify your storage architecture. Put simply, StorSimple takes all of the major elements of storage -- including primary storage, disk-based backup storage and tape infrastructure -- and puts them all on one box. From there it connects your Microsoft applications servers directly to cloud storage providers to deliver additional storage. [Also see: "StorSimple emerges from the clouds"]
In all, StorSimple promises to slash your storage infrastructure by as much as 80% and to connect you with your preferred cloud provider without any disruptive changes to your own on-premises apps.
CEO Parikh says that his company's product is the only hybrid cloud storage appliance of its kind to be certified by both Microsoft and VMware, and that StorSimple now supports cloud services from Microsoft, AT&T and Amazon, among others. What's more, it is working on supporting OpenStack.
"We provide the only solution that has core enterprise features such as high availability, no single points of failure and no downtime for maintenance," he says. "Our customers have been very happy shifting their money to StorSimple and the cloud and away from traditional storage."
StorSimple added $10.5 million in Series C venture funding led by Ignition Partners to its coffers over the summer, bringing the total to more than $31 million.
Founded: 2008 Focus: Storage for virtual machines Location: Mountain View
Who's the boss? Kieren Harty, the co-founder and CEO, worked at VMware for seven years as its executive vice president of R&D. Co-founder and Architect Mark Gritter was formerly a staff engineer at Sun Microsystems.
The lowdown: Tintri wants to be the go-to storage service for virtualized environments, as its VMstore appliance was designed only to work on virtual machines. Chris Bennett, Tintri's vice president of marketing, says that the company is comfortable putting all its eggs in the virtualization basket because virtual machines have now become the norm for IT instead of the exception.
"The number of virtual servers exceeded the number of physical servers for the first time in 2009," he says. "We're not general-purpose storage. What we want to do is remove the final block for enterprises that want to do full-scale adoption of virtualization."
Tintri provides 13.5TB of usable storage in each datastore and uses an architecture that delivers 99% of I/O from flash. Bennett says that one of the appliance's key features is its ability to auto-align virtual machines to improve performance. This particular problem with virtual machines occurs when a VM's guest OS and its storage function write data blocks to disk that don't automatically line up together. While this isn't a very big deal if you're working with a small number of virtual machines, it becomes a greater hindrance the larger you scale since misaligned VMs will inevitably create more demand for additional input/output operations per second. Tintri estimates that these misaligned blocks can degrade VM performance between 10% and 30%.
The company, which raised an $18 million C Series round of funding in June, solves this problem by using a VM-aware file system that it says "dynamically adapts" to the guest OS and doesn't disrupt its operations.
"Auto-alignment dramatically improves ease of use," says Bennett. "It's one less nagging issue for people to deal with."
Founded: 2010 Focus: SSD optimization Location: Boxborough, Mass.
Who's the boss: CEO and co-founder Duncan McCallum was previously the CEO and co-founder of Cilk Arts, a multicore solution vendor bought up by Intel. CTO and co-founder Qing Yang is an IEEE fellow who has published dozens of patents in the fields of storage architectures, disk I/O systems, parallel and distributed computing, and networks.
The lowdown: VeloBit isn't going to sell you expensive pieces of storage equipment. Rather, the company designs and sells software designed to improve the performance of solid-state drives (SSD) in the enterprise. And best of all, claims VeloBit CEO McCallum, it does this with as little disruption to your operations as possible.
"Our software enables customers to install SSDs without changing your applications, backup or data management," he says. "It will boost the performance of your SSD and give you a less expensive SSD."
VeloBit accomplishes this by using unique algorithms to create a caching layer with SSD memory that sits between applications and any block-based storage system, whether it's SAN, direct attached or SSD. The VeloBit caching layer comprises main memory, SSD and computation that compresses data at a block level and uses a variety of factors to determine whether or not it should be stored on SSD or traditional disk. McCallum says that these algorithms perform more complex tasks than typical caching algorithms that merely sort content based on how often it's accessed.
VeloBit uses a technique called Content Locality that looks for similar aspects within different blocks. So let's say you have three different blocks that each have four bytes in them, labeled "ABAB," XABX" and "XXBA." VeloBit's algorithms will recognize that while all of these blocks are different, they also all contain "AB" or "BA," so will compress those blocks of data prior to caching them. As McCallum says, this can make a relatively small cache "look really big" because so much data is chopped up, compressed and then reassembled when it comes out of the cache.
"We're not just sorting blocks, we're looking at the contents of the blocks, calculating signatures on pieces of the blocks and recording the signatures we see most often," he explains. "This helps us do compression in addition to doing caching."
VeloBit received an undisclosed amount of Series A funding over the summer from Fairhaven Capital and Longworth Venture Partners.
Read more about data center in Network World's Data Center section.