The Nirvanix Blog

The term “Cloud Storage” has become en vogue in the past 18 months, pushing the hype curve into the red. While industry buzz and news around this technology has increased, definitions have failed to root any sensibility and meaning as to what it really is and what can differentiate services that proclaim to provide it.

The purpose of this blog post is to show where the Nirvanix SDN fits into the Cloud Storage space and how our proprietary technology makes us the best choice for the enterprise looking for an alternative to building and owning a local or global storage infrastructure to support such data-intensive operations as online archive, remote backup or media library management.

A Cloud Is White, But So Is A Whiteboard

The terms “Cloud Storage” or “Cloud Computing” come from us whiteboard aficionados of the 1990s who loved scribbling a crude fluffy cloud to represent the wide area network, which in most cases now means the public Internet. To that “cloud” we would draw a line to a box representing a server connecting to the WAN or the Internet.

Today, most often, storage and computing industry professionals refer to offering these storage and compute resources over links through the Internet using web-services protocols, thus the common terms, Cloud Storage and Cloud Computing.

Cloud Storage Is Not Online Storage

At Nirvanix, we draw the clear distinction between Cloud Storage and online storage. Cloud Storage is a platform behind an Application Programming Interface (API) upon which one may build a limitless number of applications that harness the platform as its storage repository. Both Nirvanix’s Storage Delivery Network and Amazon’s Simple Storage Service (S3) are examples of Cloud Storage.

Online storage is a fully integrated application, typically singly purposed, designed and deployed to fulfill a particular storage purpose as a service. The storage component is integrated into the front end of the application in such a way that they are only together within the product. An example of online storage is EMC’s Mozy Backup Service.

Since Nirvanix is a Business-to-Business Cloud Storage provider, we enable businesses to offer their services in the market. As such, Nirvanix’s Storage Delivery Network is used commonly to create online storage products. Prime examples of this are www.wizzdrive.com or www.freedrive.com

A Cloud Is Not A Cloud: Why Some Clouds Bring Rain

Looking past all the words, Cloud Storage must be a group of physical storage servers and other supporting hardware, with virtualization software, enabling the multi-tenant partitioning of resources to store data at some point. Differentiation begins to show itself quickly here though. Most services we have reviewed to date, be they online storage or Cloud Storage offerings, tend to have some common shortcomings. At the highest level, data may be backed up but only reside online at one facility on the planet; or perhaps at a second one that purely serves for disaster recovery purposes. Other services have developers coding to two or more APIs for their different locations.

These issues pose a few problems. First, and most obvious, is that there is a single point of failure on the network. What if a backhoe digs up the fiber connects to the hosted data center and the “cloud” turns into condensation? Downtime would occur for much longer than a typical corporation would care to tolerate. There have been many dramatic examples of this in the 2008 in the cloud industry due to several causes, the backhoe excepted thus far.

The second, slightly less obvious problem, is that if you are a global business using this service, your offices proximate to these one or two data center operations will have excellent data services while more dispersed offices will not due to the physics of space and time.

Nirvanix designed a way around this by building a multi-layered, virtual file system, dubbed the Internet Media File System™ (IMFS), which unifies the many Nirvanix storage nodes located across North America, Europe and Asia into one, load-balancing storage network. Users set their availability policies in such a way that if downtime is not an option, backhoe of not, their data will be live in two or more synchronized locations around the world. Furthermore, their Tokyo and New York offices will have the same, consistent, speedy services.

To achieve these attributes, Nirvanix had to build certain components into its own brand of Cloud Storage through IMFS to eliminate these challenges presented by competing architectures.

The first component is Global Virtualization. This is perhaps the most important characteristic of a Cloud Storage service. When storing data, the namespace is what directs access to or from a file. When a storage system is installed or moved into traditional storage architectures the namespace needs to be modified to support the change. This can be a monumental task for large storage upgrades or migrations. “Storage as a service” offerings provide a virtual namespace that usually only map to one geographic location. This means you still need to develop logic to route your data to different geographical locations (if this is even an option) based upon a certain set of criteria. This also means that there is one point of failure, so in the event of downtime at the data center, the availability of the content stored in that center is compromised. Nirvanix has solved this issue by utilizing the patent-pending IMFS, which routes data to one of the many globally dispersed storage nodes based upon a user’s geographic location. Additionally, the IMFS supports policy-based data replication that enables the replication of data in up to three geographic locations.

This leads to our next characteristic, Continuous Availability. As mentioned previously Nirvanix offers the only Cloud Storage service with automated movement of content throughout the network. This means the Nirvanix Storage Delivery Network enables the most efficient storage and delivery of data regardless of a user’s location. This, combined with Nirvanix’s policy-based file replication, means that you can choose the availability characteristics that suit your business’ needs. This approach allows us to provide industry-leading service level agreements from 99.9 – 100 percent.

The third Cloud Storage characteristic is Unlimited Scalability, driven by both hardware and software. This basically means that the service provider can handle any amount of data you need to store. The IMFS also plays a significant role here allowing tremendous scale, of up to 1,000s of exabytes under a single namespace. We make scalability even easier by offering integrated application and child account management, meaning you can have multiple applications under a single account and each application can have millions of accounts under it. We not only offer unlimited scalability, we do so in a way that allows for easy integration. Nirvanix also offers the most choices to interface with the Nirvanix Storage Delivery Network via a standards-based API, an FTP Proxy and the Nirvanix CloudNAS (software that maps Nirvanix as a drive for Windows or Linux).

The final characteristic of a true Cloud Storage service is that it must be a Usage-Based Service, meaning you only pay for the services used. This is a very beneficial characteristic because companies save on purchasing hardware and software (CAPEX) while also saving on hiring additional system administrators (OPEX). Additionally, you benefit from the advantages of Nirvanix’s global network. Even if you were to continue to expand your own storage or purchase virtualization software, expanding to multiple geographically dispersed nodes would be cost prohibitive. Conversely not expanding globally limits your scalability and availability.

As you can see, being a true Cloud Storage service means more then simply offering storage through an Internet connection. It should allow you to write to one location regardless of where in the world you want your data stored. It should offer methods to guarantee the continuous access of data, eliminating bottlenecks. It should have the capacity and provide interface methods that allow you to expand in the manner that you choose. It should also provide this without requiring the purchase of hardware and do so in a convenient usage-based service model. So, while the term “Cloud Storage” may be used a little too often these days, in the press, blogs and beyond, the storage model being mentioned is not part of the true Cloud Storage model in Nirvanix’s definition, unless said service utilizes these four components. There may be a lot of “clouds” out there but not all clouds are created equally, some may actually bring rain.

Patrick Harr is part of a star-studded lineup of speakers at SYS-CON's Cloud Computing Expo. Between them, they'll be covering every aspect of the hottest IT topic for years, with not just Amazon but also IBM, Microsoft, Google, Yahoo, Intel, HP and a host of others all offering, using or developing high-end computing services typically described as “cloud computing” - through which massively scalable IT-related capabilities are provided as a service using Internet technologies.

Forrester Research analyst James Staten calls cloud computing "classic disruptive innovation - where the mainstream dismisses the product and small companies have time to create a real differentiated value." But there are so many offerings just now that what infrastructure architects are looking for above all is a set of organizing principles they can use to guide them in choosing between them all.

Such principles. and a host of associated topics, will be addressed in San Jose by a Top Speaker Faculty that includes:

• Dr Werner Vogels - VP & CTO, Amazon.com (Keynote)
• Mike Feinberg - Senior Vice President, Cloud Infrastructure Group, EMC
• Rob Weltman - Director of Grid Services, Yahoo!
  
• Peter Nickolov - President CTO, 3Tera
• Kevin Haar - President & CEO, Appistry
• Songnian Zhou - Co-Founder & CEO, Platform Computing
• Patrick Harr - CEO, Nirvanix
• Gerrit Huizenga - Cloud Solutions Architect, IBM
  
• John Keagy - CEO & Co-Founder, GoGrid/ServePath
• Sajai Krishnan - CEO, ParaScale
  
• Reuven Cohen - Founder & Chief Technologist, Enomaly
• Mike Eaton - CEO, Cloudworks
• Jonathan Bryce - Founder at Mosso (Rackspace)
• Nati Shalom - CTO, GigaSpaces
• Don MacAskill - CEO & Chief Geek, SmugMug
• Billy Marshall - Founder & Chief Strategy Officer, rPath
• Dr Thorsten von Eicken - CTO & Founder, RightScale
• Jonathan Pyke - Chief Strategy Office, Cordys
  
• Jason Stowe - Founder & CEO, Cycle Computing
• David Young - Co-Founder & CEO, Joyent
  
• Dave Durkee - Founder, CEO & Technical Director, ENKI
• John Janakiraman - CTO, Skytap
• Stuart Charlton - Chief Software Architect, Elastra
• Lars Leckie, Principal, Hummer Winblad Venture Partners
• Dr Rich Wolski - Professor, U.C. Santa Barbara
• Javier Soltero - Co-Founder & CEO of Hyperic
• Omer Trajman - Director of Field Engineering, Vertica Systems
  
• Rachel Chalmers - Senior Analyst, Enterprise Software, The 451 Group
• Ken Gardner - Executive Chairman, SOASTA
  
  and with a special Cloud Bootcamp led by
  
  
• Alan Williamson, Founder at Blog-City.com, Creator of BlueDragon
  

All breakout sessions are all listed in further detail here.



Speaker Faculty - SYS-CON's Cloud Computing Expo 2008 West


Register Today for Cloud Computing Expo and Save $200 !

Sponsorship and Exhibit Opportunities
Sponsorship and Exhibit Opportunities Offered on a First-Come First-Served Basis. To inquire about sponsorship and exhibit opportunities please contact Carmen Gonzalez at 201-802-3021 or by email at events(at)sys-con.com. Currently, a limited number of sponsorship and exhibition packages with multiple sponsorship discounts are available for the upcoming events.