GigaOM » scale-out storage

Inktank launches to change the face of open-source storage

Derrick Harris — Thu, 03 May 2012 16:17:42 +0000

The lead developers behind open-source storage system Ceph have launched a company, called Inktank, to commercialize the software. The company describes Ceph as a “fully open source, distributed object store, network block device, and POSIX-compatible distributed file system designed for reliability, performance, and scalability.” It’s uniqueness comes in part because Ceph does all these things within a unified platform.

Ceph, which was created in 2004 by Inktank Co-Founder and CEO Sage Weil while a Ph.D. student at the University of California, Santa Cruz, is already making a name for itself in the cloud computing world. Hosting provider DreamHost was an early user of the software, and has led the charge to integrate Ceph with the OpenStack project. OpenStack has its own native distributed file system, but relies on third-party options such as Ceph for block storage.

DreamHost has also worked with Dell to make Dell’s open source Crowbar configuration tool compatible with Ceph.

However, Ceph uptake doesn’t seem to have as high of uptake as other popular open source projects, so it’s hard to predict how successful Inktank will be trying to sell support and commercial services around Ceph. Hadoop, for example, had proven very popular before the ecosystem of commercial Hadoop companies popped up, and even Nginx had a huge footprint to point to before commercializing its open source web server. Ditto MySQL.

Part of the reason for Ceph’s relative lack of hype might be the presence of other popular open source distributed file systems such GlusterFS. Red Hat bought Gluster, the commercial entity behind GlusterFS, last year.

But maybe that will turn out to be a good thing for Inktank. Ceph’s unique combination of capabilities, now backed by a commercial entity, could make Inktank stand out as the open source platform of choice for businesses that want to avoid the large-vendor experience.

Image courtesy of Shutterstock user Jason Mintzer.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

Red Hat buys Gluster for scale-out storage

Barb Darrow — Tue, 04 Oct 2011 13:56:57 +0000

Red Hat is buying Gluster and its open-source storage know-how. The acquisition will give Red Hat a strong play in the scale-out file system space as it steps up competition against incumbent IT vendors to host the next generation of enterprise applications.

In a statement, Red Hat CTO and VP of worldwide engineering Brian Stevens said:

Our customers are looking for software-based storage solutions that manage their file-based data on-premise, in the cloud and bridging between the two. With unstructured data growth (such as log files, virtual machines, email, audio, video and documents), the 90′s paradigm of forcing everything into expensive, single-system DBMS residing on an internal corporate SAN has become unwieldy and impractical.

This $136 million cash deal is just the latest instance of a big vendor buying its way into the hot scale-out storage market. Last year, for example, storage power EMC bought Isilon for its scale-out expertise. Before that, Hewlett-Packard bought Ibrix. IBM’s purchase of Storwize and Overland Storage’s acquisition Maxiscale were all part of this cloud-inflected, scale-out storage land grab, too.

In the pricier, traditional scale-up model, lots of new nodes are piled up behind a few big, pricey controller servers. In scale-out storage, companies can add more inexpensive commodity server nodes horizontally, and such additions tend to be relatively easy and inexpensive.

Red Hat said it will continue to sell and support Gluster’s existing products, and it will also incorporate Gluster technology into other Red Hat solutions. It will sell Gluster’s services via Red Hat’s usual subscription model. Sunnyvale, Calif.–based Gluster claims 150 customers, including streaming music pioneer Pandora.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

Can the mid-market drive an infrastructure revolution?

Derrick Harris — Tue, 16 Aug 2011 21:00:09 +0000

Nutanix today announced the general availability of its converged infrastructure appliance, and CEO Dheeraj Pandey is counting on mid-size enterprises to drive its adoption and that of similar products. His stance is that SMEs have the budgets, the IT demands and the right cultures to justify taking chances on new approaches to infrastructure. He could be onto something.

With its Complete Cluster, Nutanix is trying to revolutionize enterprise storage by eliminating the SAN and placing both computing and storage on the same node. Nutanix has referred to this as a Google-like infrastructure for mid-size companies, at least in the respect that it collapses the traditional three-tier application architecture into a single appliance. The 2U cluster, which targets virtualization workloads, includes four nodes consisting of 8 6-core Intel processors, up to 768GB RAM, 1TB of Fusion-io solid-state storage, and a combined 35TB of hard drive capacity.

The converged architecture represents a pretty big change of direction for many IT departments, but storage — historically and presently — has been an area where mid-size companies are willing to embrace disruptive technologies. That’s because, Pandey said, mid-market companies have respectable IT budgets but are “chronically understaffed,” so they just want technologies that get the job done and they’re willing to pay for simplicity. Mid-size companies don’t have any religion when it comes to vendors or particular architectures, he added.

They’re also struggling with a higher percentage of virtualized workloads, which can lead to significant performance overhead within storage systems. Trying to do take advantage of virtualization to do dynamic orchestration and other cloud computing-like tasks adds an even greater burden on legacy infrastructure that was designed to run physical workloads. Flash storage and management software designed with virtualization in mind can help alleviate some of these woes.

Pandey points to iSCSI as one example of mid-market companies bucking the industry trend (toward Fibre Channel) and choosing the option that worked best for them. Lately, one might also look at the success of flash-based storage startups such as Nimble Storage, Nimbus Data Systems and Tintri. All these companies are having considerable success selling appliances that disrupt traditional storage notions by replacing spinning disks with solid-state drives and targeting virtual workloads, and they’re selling to mid-size enterprises.

Of course, Pandey thinks Nutanix has an advantage over companies selling storage appliances only because it doesn’t have to sell to storage administrators who will always be comparing new technologies against NetApp and other legacy vendors’ technologies. He says Nutanix is selling to higher-level IT folks who are looking for the best overall infrastructures on which to run their cloud and virtualized workloads. With Nutanix, the storage component is just a part of the greater appliance that includes VMware software and computing hardware.

Ultimately, though, whatever the form factor, it’s the price-performance ratio that’s driving adoption of these new architectures. Converged infrastructure is nothing new, but Nutanix’s $75,000 entry point is a far cry from the list price for a VCE Vblock, for example. Nimbus Data Systems and Nimble Storage also claim their flash-based appliances provide better performance than hard-disk-based systems while competing very strongly on price.

There is a great deal of business to be done outside the Fortune 500, and if Nutanix, Nimbus and others trying to revolutionize storage infrastructure can keep convincing customers that their products deliver more bang for the buck, we could be witnessing the beginning of a storage revolution that results is spinning disks and SANs becoming things of the past for many companies.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

Hadoop may be hot, but it needs to be useful

Derrick Harris — Tue, 28 Jun 2011 20:45:45 +0000

Hadoop is a very valuable tool, but it’s far from perfect. One potential concern for businesses whose primary products don’t end in .com is that it was built with the web in mind. That means it was designed for massively scaled architectures and to handle petabytes of data, and that implementing and managing it can require a team of Ph.D.-level engineers.

For companies that don’t operate at webscale certain aspects of Hadoop look like overkill, and getting it to address their specific needs look like Herculean chores. For these companies, Hadoop just has to be useful.

While vendors including Cloudera, EMC, MapR and Yahoo, as well as the Apache Software Foundation, are working (some might say fighting) to shape distribution-level Hadoop concerns such as the core MapReduce and file-system architectures, vendors up the stack are focusing on business-level concerns, often without regard for what’s running underneath. Two of them — Platform Computing and Pervasive Software — released new products this week.

Platform MapReduce product aims to brings that company’s expertise in distributed-systems management to Hadoop clusters. I detailed Platform’s MapReduce vision in March, and the new product is a first step toward delivering on that vision. Platform supports numerous MapReduce and file system distributions, but its application-level management features make it shine.

Probably chief among them is the ability to maximize utilization of Hadoop hardware by running multiple applications on the same cluster. With advanced scheduling features and, the company claims, “10,000 priority levels,” numerous applications can run simultaneously while Platform’s software takes care of determining which one is running where and which one can access the shared data at any given time.

Pervasive, a data-integration specialist that has turned its attention to Hadoop, takes a different approach to catering to enterprise customers. For a while now, it has been selling a product called DataRush, which lets users write MapReduce workflows optimized for multicore processors, resulting in faster performance on fewer nodes. Yesterday, it announced early access for a product called TurboRush, which brings those same capabilities to Apache Hive, the Facebook-created tool for bringing traditional SQL database and data warehouse features to Hadoop clusters.

The premise behind DataRush and TurboRush is compelling because although it’s fascinating to talk about Hadoop clusters at Yahoo and Facebook that span tens of thousands of clusters and store petabytes of data, most businesses don’t have those webscale needs. As I noted in a recent GigaOM Pro report on Hadoop (subscription required), the average cluster size among Hadoop users appears to be less than 50 nodes and nowhere near petabytes of data. There’s no use buying, managing or paying the power bill for more servers than necessary if a relatively small group of multicore processors could do the trick.

With Yahoo’s Hadoop Summit taking place tomorrow, we’re destined to see lots of news about new distributions, tools and services. And although distributions and architectural innovations rightfully steal much of the spotlight, it’s all important. Neither Platform, Pervasive nor any other vendor has all the answers when it comes to Hadoop, but the good news is that whatever your specific concern, there’s likely someone either selling or working up a solution to it.

Image courtesy of Flickr user MrFarber.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

Nutanix Gets $13.2M for Google-like Storage Architecture

Derrick Harris — Tue, 19 Apr 2011 14:12:27 +0000

Nutanix, a San Francisco-based storage hardware maker, has raised $13.2 million in a Series A funding from Lightspeed Venture Partners and Blumberg Capital. The company is developing an appliance combining computing and storage on the same server nodes, a story that should resonate with customers concerned with scalability and performance.

When it comes to scale-out architectures and appliances, Nutanix’s founding team knows from whence it speaks. Co-founder, President and CEO Dheeraj Pandey was VP of engineering at Aster Data Systems and designed the storage systems for Oracle Database and the Exadata appliance before that. CTO Mohit Aron was a lead architect at Aster Data after spending time at Google designing the Google File System. Co-Founder and Chief Products Officer Ajeet Singh also worked at Aster Data and previously helped develop Oracle’s cloud computing strategy.

Pandey compares the Nutanix appliance to Google’s architecture in that computing and storage are both house in the same nodes. This is different from many traditional application architectures, where computing is housed on one set of servers and storage is either network-attached or housed in a separate SAN. With the Nutanix approach, Pandey explains, both storage and computing scale simultaneously, which can lead to massively parallel processing and storage. A big differentiator from other scale-out storage products, he said, is that many are file systems, but Nutanix is not so limited. The Nutanix appliance includes data management software that is akin to “bringing all of NetApp” to the system.

The key to Nutanix is virtualization, which provides the abstraction and the additional storage connections necessary to give Nutanix the performance edge it claims. The company is big on solid-state drives for performance and consolidation, but Pandey says legacy storage systems are limited to the amount of SSDs they can handle. With a virtualized computing layer, however, each virtual server and each physical node provide the requisite housing and connection to an additional SSD. The Nutanix appliance combines both SSDs and hard disk drives to achieve maximum levels of performance and affordability, Pandey said.

Despite the known difficulty of selling appliances versus software alone — something we’ve seen played out recently by both Schooner and Cirtas Systems — Pandey is confident in Nutanix’s chances. For one, he noted, converged infrastructure is hot right now thanks to products such as Cisco’s Unified Computing System, the Virtual Computing Environment’s Vblocks, HP’s BladeSystem Matrix and Dell’s vStart. However, explained Co-Founder and Chief Products Officer Ajeet Singh, those products involve separate storage components like customers could buy separately; the systems are really just enclosures. Pandey says it comes down to a choice between choosing the iPhone approach to integration or the Android approach of software on multiple devices, and Nutanix chose the former.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

When It Rains, It Pours (Cash) for Storage Startups

Derrick Harris — Sat, 05 Mar 2011 01:05:11 +0000

It seems like when it rains for cloud storage startups, it pours, and this week was no exception. Egnyte closed a $10 million round led by Kleiner Perkins Caufield and Byers, Scality closed a $7 million round led by IdInvest Partners, and RightScale chose Gluster to provide scale-out NAS within RightScale’s cloud management platform. The concept of big data (i.e., storing and analyzing large amounts of data) gets a lot of press — in fact, we’re putting on an entire event around it — but the huge amounts of money being poured into storage companies utilizing cloud architectures prove there’s still plenty of demand for fast, cheap and scalable file storage.

Since I covered three startups scoring a combined $30 million in three days’ time in November, the spending and funding has only picked up. SolidFire scored $11 million, Nasuni got $15 million, Scale Computing closed $17 million and Cirtas stole the show with $22.5 million. And don’t forget Isilon, the clustered file system vendor that sold to EMC for $2.25 billion. For Gluster, which contributed $8.5 million of that three-day splurge in November, RightScale is its second big-name, cloud-based customer win of the new year, as it announced Pandora as a customer in January.

With so many cloud storage startups flying about, I’d be lying if I said it isn’t getting difficult to keep track of which one does what. But I don’t think I’ll have that problem for long. Just like consolidation started among cloud computing startups over the past several months, cloud storage consolidation can’t be far behind. Then, the problem will be keeping track of who bought whom, and for how much. Still, I suspect a bit more investment before that happens, but there are a handful of startups in this field that appear too valuable to stay independent for long.

Related Content from GigaOM Pro (sub req’d):

Will Streaming Content Drive Scale-out Storage?

Derrick Harris — Wed, 05 Jan 2011 17:30:50 +0000

It has been a good couple of months for open-source clustered file system vendor Gluster. In November, the company closed $8.5 million in investment funding; Wednesday, it announced streaming music pioneer Pandora as a customer. The news validates the billions that large IT vendors and investors have poured into scale-out storage providers — and also because it suggests streaming content providers are a possible target market for these providers going forward.

Streaming content is only getting more popular, as evidenced by incredible amount of bandwidth currently — and predicted to be –consumed by people watching video online. According to the press release, Pandora has deployed 250 terabytes of storage to manage its growing library of song files, and each requires multiple versions to enable streaming to numerous devices types. It seems logical that a company like Apple — with huge music and video libraries, and a seemingly strong desire to stream it — needs even more storage capacity. As Amazon CTO Werner Vogels noted when Amazon Web Services said that it now supports 5TB objects in Amazon S3, HD video files can be terabytes apiece. Assuming the list of services streaming consumer content continues to grow, they’ll all need file systems that can handle their mountains of unstructured data.

Of course, Gluster is only one option (albeit unique because of its open source nature) in a sea of options, and a relatively small one at that. EMC recently paid $2.25 for Isilon; NetApp now owns Bycast’s StorageGRID technology; and Hitachi recently absorbed ParaScale. But if streaming services continue to proliferate, there should be enough business to go around.

Related content from GigaOM Pro (sub req’d):

Storage Funding Marches On: $17M for Scale Computing

Derrick Harris — Tue, 23 Nov 2010 17:00:02 +0000

Fresh off a $9 million Series B Round in March, clustered file system startup Scale Computing has closed a $17 million Series C round. Led by Scale Venture Partners, along with Northgate Capital and Benchmark Capital, the round brings Scale Computing’s total funding to $31 million. It’s yet more proof that spending on scale-out storage software is nowhere near its end.

Scale Computing Intelligent Cluster Storage is not unlike the file-system software offered by Isilon Systems and Gluster, both of which have done alright for themselves lately. Gluster raised $8.5 million earlier this month, and EMC bought Isilon last week for $2.25 billion. All of this is just the latest in a year’s worth of heavy investment in scale-out storage, through either venture capital or acquisition.

The connection to cloud-based storage — which also is white hot lately — only helps the prospects for future investment. Scale-out file systems work both on-premise and as the foundation for cloud-based storage platforms, so even after storage vendors all have their strategies set, up-and-coming cloud providers will still be in the market. The converged infrastructure trend, too, is driving many server makers to consider a storage play, and products like those from Scale Computing, Gluster and others are ideal for scale-out commodity server environments.

Image courtesy of Matthew Field.

Related content from GigaOM Pro (sub req’d):

Cloud Storage: Two Days, Three Startups, $30 Million

Derrick Harris — Wed, 10 Nov 2010 20:57:45 +0000

Cloud storage startups have received lots of investor love this week. In the past two days, three cloud storage companies raised a total of $30 million, with Gluster bringing in $8.5 million, Zetta scoring $11.5 million and and Nirvanix receiving $10 million. It’s further proof that organizations are looking for scalable (and, ideally, inexpensive) methods for storing their growing data stores, and that cloud computing is becoming an ideal model to do that.

This week’s funding announcements aren’t the beginning of any trend, merely a continuance. All told, Gluster, Zetta and Nirvanix have raised a disclosed total of $75.7 million, with Nirvanix leading the way with $45 million. Zetta’s new money makes it the latest addition to the $20 million club; the company has raised $22.5 million overall. Both Nirvanix and Zetta target enterprises with cloud-based backup and archiving offerings.

Further, the past year has brought some entirely new cloud-storage startups to market. In December 2009, Nasuni announced an $8 million initial round, and brought its cloud file system to market in February. In September, Cirtas Systems exited stealth mode with $10 million and its Bluejet Cloud Storage Controller. Nasuni targets SMBs looking to manage primary storage in the cloud, whereas Cirtas targets enterprise retention with an appliance that intelligently directs data to the cloud.

Proprietors of cloud, or scale-out, storage software — which is where the open-source Gluster plays — have had a good year, too. In March, Om highlighted a number of big scale-out storage investments. In April, NetApp continued a trend of scale-out acquisitions by purchasing Bycast. In the past few months, IBM bought Storwize and Overland Storage bought MaxiScale. Lately, talk has been of Isilon Systems marketing itself for a 3PAR-style auction (although it appears initial frontrunner EMC is out as the buyer). Scale-out file systems, in particular, are popular among individual organizations, as well as cloud providers building scalable, multi-tenant storage infrastructure.

Financially speaking, the only big loser in cloud storage appears to have been one-time VC darling ParaScale. After raising $11.4 million in 2008, ParaScale couldn’t raise additional funds, resulting in Hitachi Data Systems saving ParaScale from bankruptcy by buying the company in August. It’s tough to assess why ParaScale got shut out of additional money, but it might have something to do with the sheer amount of competition.

With all this money flying around cloud-storage platforms and software, it’s difficult to speculate who’ll get bought or who’ll get the next big injection of venture capital, but it appears certain the action isn’t slowing down.

Related content from GigaOM Pro (sub req’d):

Doubling Down on Scale-out Storage

Gary Orenstein — Sat, 10 Apr 2010 16:00:05 +0000

NetApp this week agreed to acquire Bycast, whose storage virtualization is used for large-scale digital archives and storage clouds — yet another investment by a major systems provider in scale-out storage aimed at tackling the growth of unstructured data. But while big systems vendors realize they need a new approach to solve the workloads generated by the web, cloud and data-intensive applications, are they really ready to ride the commodity hardware cost curve embraced by large web and cloud providers?

Scale-out storage differs from traditional scale-up architectures in that it horizontally adds new storage nodes (servers with internal disks) instead of vertically aggregating lots of individual disk drives behind one or two super-sized controller servers. Scale-out further implies a corresponding software layer that can make hundreds or thousands of nodes act like a single system. It has the potential to change the shape of the data center by displacing large, costly enterprise equipment with an easily sized pool of inexpensive servers.

Google and other web giants have found great success with scale-out approaches. Indeed, for web and cloud computing workloads, increases in users, objects, and capacities drive a need for innovative scaling approaches at low costs, and scale-out storage, by leveraging commodity hardware, fits the bill. (Data storage is just one of the issues we’ll be taking an in-depth look at June 23-24 at the GigaOM Network’s Structure conference in San Francisco.)

The race to capitalize on high-growth, scale-out data markets has led NetApp, HP and EMC to spearhead acquisitions and new product development in this sector. In 2003 NetApp also acquired Spinnaker, which at the time was positioned primarily as a way to aggregate multiple file systems into a single larger file system, or what’s known as a single global namespace. The Spinnaker product resulted in a “tortuous integration effort” according to some, and obviously didn’t fit all of NetApp’s needs when it came to providing scale-out storage, hence the Bycast buy.

HP, similarly, has been around this block a couple of times before. In 2007 the company acquired Polyserve, a clustered file system that had integrated with Oracle solutions, and then in 2009 bought IBRIX, a scale-out storage provider with solid penetration in the animation arena.

EMC, meanwhile, has also had multiple product plays at work. Its first effort to tackle these new scale-out object-based workloads was Centera, which it positioned as a content-aware storage solution. More recently, EMC’s been working on Atmos, a cloud storage product aimed at turning service providers into purveyors of terabytes online.

For NetApp, one of the appealing features about Bycast’s offering is that it doesn’t necessarily replace the underlying storage systems, but rather aggregates existing solutions, including those by NetApp. IBRIX, which was once adopted by customers on a range of hardware platforms including a long relationship with Dell before the HP acquisition, now seems to be tightly packed with HP hardware. And EMC’s offerings do not come in a software-only flavor.

It seems like the appetite to reach the scale-out storage market is still carefully balanced with the existing product lines and systems that form the lifeblood of big vendor sales. But these moves also signal a recognition that scale-out is here to stay, and that NetApp, along with other major systems vendors, will not stop at one acquisition or product to capture the market.

Related content from GigaOM Pro (sub req’d):

What’s Next for the Cloud? Distributed Architectures

Gary Orenstein is host of The Cloud Computing Show.