"Cloud computing is here to stay as a major IT wave that will take its place in the data center alongside mainframes, client-server and 3-tier web, says Sajai Krishnan, CEO of ParaScale, in this Exclusive Q&A with SYS-CON's Cloud Computing Journal.

"The recession is driving IT to consider newer, lower cost offerings and cloud computing is delivering on these requirements," he adds. Krishnan specifies the two primary concerns for users of the cloud right now to be Security and Lock-in.

Cloud Computing Journal: David Linthicum, the industry analyst and commentator sees Cloud Computing as being divisible into ten distinct patterns:

• Storage-as-a-Service
• Database-as-a-Service
• Information-as-a-Service
• Process-as-a-Service
• Application-as-a-Service
• Platform-as-a-Service
• Integration-as-a-Service
• Security-as-a-Service
• Management/Governance-as-a-Service
• Testing-as-a-Service

Do you think it is just coincidence that he lists Cloud Storage first? Or does Cloud Storage undergird the entire stack?

Sajai Krishnan: It is no coincidence that storage is listed first. With any system, storage is the persistent foundation for which the data starts and finishes. Cloud storage is no exception and can even introduce unique challenges when compared to cloud computing. Take the classic use case of cloud computing:

My company needs to spin up 2000 servers in one day to deal with unforeseen demand or 20 servers at the close of each quarter. Two days later I spin them down when the demand passed. Cloud computing saves the day.

Given the need for an organization to save everything, storage growth tends to be steady and relentless. Shrinking your cloud storage footprint after a demand event is more difficult and less common (outside of special situations like simulations, data mining, scratch storage situations, etc.

Cloud Computing Journal: Are there problems associated with Amazon and Google's cloud computing concepts - to take tow leaders in the space?

SK: There are two primary concerns for users of the cloud: Security and Lock-in.

Security: Shared infrastructure scares many enterprise customers. How do they know their data is safe or even in the specified location? There have been numerous articles published about this issue so I'm not going to dive into the details but the concern is real. Combined with privacy concerns, this issue will be something that will continue to dog public cloud services.

Cloud lock-in: Each public cloud vendor has created a unique interface or API for using their offering. A user must program to that specification. If the user wants to change providers, they must reprogram to a different specification and pay double bandwidth charges for moving data. This puts the onus of change on the customer and enables cloud lock-in for the vendors.

Cloud Computing Journal: How does ParaScale's approach circumvent those problems?

SK: ParaScale addresses these issues by enabling the enterprise to build a private cloud and write files using standard protocols. We do not require a custom language to "speak ParaScale", instead we support industry standards like NFS, FTP, and HTTP that are well understood and widely deployed (in addition to new protocols like WebDAV).

A private cloud is deployed inside a customer's firewall and is managed by internal employees. Established security practices can be used to secure data and access yet the enterprise still benefits from the scale and economics of cloud storage. Additionally, with the ability for both service providers and enterprises to leverage ParaScale technology, before broad-based standards are established, end-users can transition between public, hosted and private clouds.

Cloud Computing Journal: What are the main new security issues that IT needs to address when storing data in the Cloud?

SK: Control of data is the main new security issue. Security is a broad subject but when considering public clouds data control is new and should be considered. There are plenty of unanswered questions about data control and reporting responsibilities. If a public cloud service provider is subpoenaed by the government it could well be that your data is vacuumed up, even though your company is not being specifically subpoenaed.

Think about what we have seen in the cell phone industry. Even more unclear is who is responsible for the data. Some argue that control is based on locality and therefore the cloud provider is responsible. Others say control lies with the owner of the data. Until this is feted out in the courts, this will be an open issue.

Cloud Computing Journal: How does the Public vs. Private paradigm apply to storage provisioning in the Cloud?

SK: In general, both public and private provisioning decisions come down to capacity and quality. Public clouds offer one model for provisioning storage. Private clouds have the ability to provide several options such as thin provisioning, fixed provisioning or a combination of those two.

Cloud Computing Journal: So companies actually can choose themselves whether to introduce cloud storage inside or outside the corporate firewall?

SK: Yes. In fact we expect many customers to choose both. Highly sensitive data will be kept on a private cloud. Data that needs geographic deployments and is not sensitive can be put into the public clouds. Others will use the public cloud to backup a private cloud with their service provider. New options are emerging every day, and end users benefit from selecting the best option based on cost and security.

Cloud Computing Journal: How about cost? Do you agree with those who contend that the lowest cost per gigabyte or terabyte wins?

SK: No. Cost is more than a purchase price. If you sell something for a penny a TB but it takes 15 guys to manage it, it's a bogus solution. If the user has to re-write applications and break security practices to save money, it will not fly. Cloud drives new economics that save the enterprise money from both a CAPEX and OPEX standpoint. Entry prices are lower due to commodity hardware and standard protocols. OPEX is saved via automation and policy based management, removing the need for multiple administrators per petabyte.

Consider, for example, the issue of data migration. NAS implementations require downtime and significant effort to move data from device A to device B. Usually a weekend or two is involved and end users need to be remapped to new mount points. The disruption is significant enough to enable a new class of file virtualization appliances that sell for hundreds of thousands of dollars each. With cloud storage, data migration is automatic and doesn't require weekends (or administrator interaction for that matter). This is one example of the exponential simplicity the cloud architecture enables.

Cloud Computing Journal: Does ParaScale pricing conform to the Cloud Computing norm of "pay-only-for-what-you-eat"?

SK: Yes. Even better, with ParaScale you can provision virtual file systems larger than your physical capacity, enabling your eyes to be bigger than your stomach. As files are written to the cloud, the administrator can add capacity just in time, leveraging newer hardware at a lower cost and larger capacity. And acquiring the standard Linux servers is really easy versus having to wait for the traditional specialty storage appliances.

Cloud Computing Journal: And scalability, presumably, is all taken care of?

SK: Others talk of unlimited scalability but everything has limits. No system is "infinitely scalable" and claims as such are bogus. ParaScale is designed to scale to hundreds of nodes and multiple petabytes. For customers who want to go bigger than that we recommend a few clouds connected logically with a single global namespace. It's not an architecture limitation; instead it is a practicality of networking capabilities and testability.

Cloud Computing Journal: Aside from storage, is there any other aspect of the Cloud Services stack that ParaScale has identified as having growth potential for the company?

SK: ParaScale is focused on cloud storage. Our roadmap is filled with enhancements that will enable new management paradigms, simplify application integration and continue to drive economies of scale.

Cloud Computing Journal: Last summer you raised $11.4M in a funding round; how have those funds mainly been applied to date?

SK: Our core technology has been running in customer environments for many years and proven itself stable and reliable. At GA, we plan to make our code available via a web download to anyone who registers on our website. Users can build a cloud up to 4TBs in size and use it forever, free of charge. We don't ask you to believe our marketing, instead try the software for yourself. To enable this model the software must be as simple as installing an application on your laptop. Therefore since funding we have been focused on making our product simple to install and configure.

I can personally attest to the simplicity as I've downloaded and built a cloud myself in an afternoon, and remember, my specialty is not systems administration.

Cloud Computing Journal: What's the risk of Cloud-* becoming just another buzz phrase used in the attempt to get organizations to "sign a check"?

SK: In this economy you need more than a buzz phrase to get a check cut. It's true that every vendor is jumping on the cloud bandwagon and trying to mould the definition to fit their offering, ParaScale included. But customers will look for offerings that solve their challenges regardless of the tag applied.

The cloud tag may get vendors in the door, but without the architecture to back up the claim, customers will move on. With any new hyped technology there is burst of marketing "me-too". Cloud is no different and as with new areas there will be a shake-out. But beyond that, it is our view that cloud computing is here to stay as a major IT wave that will take its place in the data center alongside mainframes, client-server and 3-tier web.

Cloud Computing Journal: Do you agree with those who contend that Cloud Computing, in its current incarnation anyway, falls somewhat short on its promise to make computing as a whole as simple as plugging an application into a utility service?

SK: Cloud computing is in its infancy and the protocols and implementations are not ubiquitous (yet). It's a bit like an American running around the UK trying to find a 110V electrical plug. The answer is always, you need an adapter. When a standard emerges and any application can talk to any cloud the promise will be delivered.


Cloud Computing Journal: What's the future trajectory of Cloud Computing seen from a ParaScale perspective?

SK: Upward and to the right. Cloud Computing is just getting started. Early adopters are getting on board and helping the industry make better products. The recession is driving IT to consider newer, lower cost offerings and cloud computing is delivering on these requirements. At ParaScale our early customers are excited about cloud storage and dreaming up use cases we haven't considered. We are on the cusp of the next wave of computing and at ParaScale, surf's up.