Q&A: The Truth about Private Clouds
Truths and myths about private clouds.
Private clouds are catching on, but IT must carefully consider how to introduce them into an enterprise. We explore the common mistakes and oversights IT makes when evaluating the technology, how to choose what data should reside on private clouds, and where this storage option is headed.
To learn more, we turned to Sajai Krishnan, the CEO of ParaScale, a cloud storage software provider.
Enterprise Systems: There has been a lot of buzz about cloud computing. Why is all the emphasis on public clouds. Why aren't private clouds more popular?
Sajai Krishnan: All things cloud grabbed headlines during the past year. Debates raged about its definition and whether "private" clouds even existed. Most recently, Amazon entered the space with its own virtual private cloud offering, adding to the private cloud emphasis of vendors including VMware, Sun, etc. Although still a hosted cloud, Amazon helped to validate the concept of a private cloud and confirmed the value businesses place on security and privacy. Understanding of the cloud is still nascent in the market. In ParaScale Webinars, 60 percent of attendees still attend to "figure out what the hype is all about." This is not surprising given the number of public cloud options, which will soon total about 100 or more.
Why do private clouds make sense for business?
Private clouds easy to use and save money. They address the exploding requirements for file data storage (unstructured data), especially private cloud storage which is as straightforward as NAS (network attached storage) and appears NAS-like to existing applications.
What are some of the popular misconceptions about private clouds?
Many companies don't think they are big enough to own their own cloud. "After all, isn't cloud the inter-galactic architecture popularized by Google and Amazon?" Most aren't aware that it is easy to start with a few servers and grow, as you benefit from the shared resources using commodity hardware including repurposed servers.
What are the practical considerations for introducing private cloud storage into an enterprise?
Start with a solution that requires the least amount of change to the organization. Storage solutions requiring re-coding of applications or changes to user behavior are not likely to succeed. Private cloud storage supports standard protocols such as NFS and CIFS that simplify introduction and drive cost savings from day one.
Start small and grow with demand. Unlike traditional storage solutions you do not have to buy a large, fixed set of capacity to get a competitive price. Cloud storage can start with a heterogeneous mix of repurposed servers storing a few TBs.
What are some of the common mistakes/oversights IT makes when evaluating private cloud storage?
The most common mistake is to view cloud storage as a direct replacement to tier-1 NAS or SAN storage. Cloud storage complements existing infrastructures and enables the enterprise to align the value of the data with the underlying storage technology. There are true advantages in scale and management with cloud storage (private or public), but businesses should not be planning to shift mission-critical database applications to the cloud, or look for NetApp's excellent snapshotting technology for rapidly changing data.
Another common mistake is to think that clouds are only capable of internet trickle speeds. Private clouds can operate at LAN speeds easily saturating gigE networks, and applications with multiple clients accessing multiple files can see performance at several times that of a typical Tier 1 NAS appliance. That essentially enables enterprises to save money on about 80 to 90 percent of the terabytes in a typical enterprise with cloud storage
What are the implications of private cloud storage for security administrators?
They're minimal for security administrators. Private clouds are deployed inside the firewall and leverage common tools such as Active Directory and encryption of data at rest. Private clouds also offer native multi-tenancy support for segregating users within the enterprise. This is supported natively in the architecture and eliminates the need for complex networking designs with traditional storage options.
How do you decide what data is appropriate for the cloud vs. residing on-site?
Cloud storage is designed for storing file, not block, data. A storage cloud is especially great for storing large files such as backup images, VM repositories, document archives, rich media, log files, video surveillance data, and large genomic data files.
How does private cloud storage fit into the current infrastructure, and how does it impact storage administrators? For example, are there tools available that help you manage legacy data (pushing disk files to tape after one year without access, etc.) in the cloud the way you do with your on-site storage?
Look to complement the existing infrastructure with private cloud storage. Every enterprise has data sitting on Tier 1 storage that does not belong there. A UC Berkeley study from June 2008 looked at a typical enterprise with 200+TB of data and showed that 90 percent of data is never touched after three months. Use your favorite Tier 1 storage for you new data, and put the rest in a storage cloud. There are many ILM and HSM solutions that can help the enterprise discover and move data to the appropriate tier.
As data sets grow larger, won't bandwidth become a limiting factor that prevents private cloud storage from being cost effective?
Clouds solve the bandwidth problem that traditional systems cannot. Adding nodes to a cloud increases both the overall capacity and bandwidth. Each node exports the global namespace and can service any request. As the data set grows, so does your bandwidth.
Where is cloud storage headed? How will it be different in, say, a year from now?
The availability of cloud storage software from ParaScale and EMC Atmos is leading to a rapid escalation of service providers entering the market with varied cloud storage services and CIOs deploying private storage clouds. Initial private cloud storage deployments are targeting big content farms, archival, and Tier 2 storage needs. Even in a recession, IT organizations are struggling with the never-ending growth in storage requirements. Cloud storage addresses these requirements with economy and value. This year we're seeing solid uptake from the early-adopter segments. The weak economy is going to be a catalyst for broader adoption in 2010. Why spend more money on Tier 1 storage when 90 percent of the data is not Tier 1?
Where does ParaScale fit in this market?
ParaScale is the only vendor to offer downloadable software that can be deployed behind the firewall as a private cloud on commodity Linux hardware. ParaScale takes Amazon- and Google-type cloud technology and brings it to the enterprise in an easily consumable way, using standard protocols such as NFS. Existing applications can work with a ParaScale cloud with no change. The software residing on multiple Linux servers aggregates the internal server storage to provide a highly scalable storage cloud with the ease of management of a single appliance, but with massive capacity and parallel throughput. Deployments can start small and scale to several petabytes in a single pool managed by a single administrator. Interested parties can download a free version from the ParaScale Web site (http://www.parascale.com) and build a cloud storage infrastructure.