Amitabh Srivastava, president for EMC's Advanced Software Division is the man behind the storage giant's much-talked about software-defined storage (SDS) initiatives. He talks in detail about EMC's ViPR and why SDS has been wrongly-defined so far.
You were the brains behind EMC's game-changing product, ViPR. What led to the development of this product?
Srivastava: We started our work around ViPR way back in 2011. Customer feedback was at the core of our effort. There were three important things that enterprise customers were looking at. Lowering the opex was one of their top-most priority. Enterprises today manage their data in silos, which makes the operational expenses directly proportional to the data growth, and therefore it's not a sustainable model. Secondly customers wanted more choices, in terms of getting the best of breed products to maximize the efficiency. Last but not least, most of the customers are looking at transitioning to what we call the 'platform 3'. If we go by the IDC definitions, Platform 1 was the era of mainframe, platform 2 was the era of PCs, and the client-server model. The recent 'platform 3' is the new age of cloud, social and big data. That is exactly where the industry is today and the enterprises are trying to move to. This shift from platform 2 to 3 is a very challenging phase for the enterprises. ViPR was the outcome of our efforts to solve all these customer issues in one shot.
How has been its uptake so far?
Srivastava: ViPR is a product that was built from scratch and was GA'd (generally available) on September 27, 2013 so that customers can use it and take a first-hand experience. We have sold this to customers from the largest enterprises, which would include financial institutions, service providers and others. It's been a quarter since the product was released and we have almost doubled what we had targeted for.
One of the key highlights of this product is its ability to work in heterogeneous, multi-vendor environments. What was the rationale behind it?
Srivastava: We know that real environment is going to be a combination of EMC hardware, non-EMC hardware and commodity hardware. EMC is not imagining the real world to be 'all EMC'. Enterprises are always going to go for multi-vendor support and environment. Hence we decided that ViPR should support NetApp, Hitachi and EMC among others. This year EMC is planning to support commodity hardware as well. Besides, customers want choice. If the customer wants to buy an array made by our competing vendor, he has the freedom to buy that even if they have bought ViPR. They wouldn't want to be stuck with EMC arrays. No customer would want to be in that fix; hence ViPR gives them the choice.
EMC may be against creating hardware lock-ins, but will ViPR not create some sort of a 'software lock-in' eventually. The management layer is the stickiest of all.
Srivastava: EMC is not a defensive player, but more of an offensive player. Our focus is to solve the problem. If NetApp's arrays are better for the customer, ViPR will support them. But the reason EMC is not playing defensive is because EMC sells the best arrays. In this environment, EMC lets its customer decide which system suits them. In fact, EMC has done a few things to avoid software lock-in. The APIs that connect the multi-vendor arrays are open; so anyone can take any array and can plug it into ViPR. Similarly, the APIs that connect to the multiple management systems--whether it is VMware or OpenStack--have been left open. We have even completed the work to integrate it with VMware and Openstack. The APIs on which a developer can add new capabilities are also open. ViPR is as open as it can be.
Can a customer download and use ViPR?
Srivastava: Yes. Though the product will be starting to be shipped soon, the ecosystem of the product is still under trial phase. So customers can download it for free from EMC website for non-commercial use. Apart from working with large enterprises and customers, EMC is also trying to develop an ecosystem around it. So we are working with a lot of large SIs, partners and start-ups to innovate on our platform.
So is ViPR your single answer for SDS?
Srivastava: Yes, ViPR and the new software capabilities that we keep adding in software over a period of time. EMC believes that ViPR is our SDS.
You recently mentioned that 2014 is going to be the year of SDS and software defined everything. Isn't it a bit too early for this technology to become mainstream?
Srivastava: If we look at the three components--compute, network and storage--of automation, it's pretty evident that server virtualization and Software-defined networking (SDN) are far more matured as compared to software-defined storage. With that, the word 'software-defined' became a buzz word and a marketing jargon. Everyone started saying that everything that they were doing is SD. Everyone is coming up with their own spin to it. Customers got inundated by all this. But I believe customers will eventually be made to sit up and take notice of SDS, as the data is growing at a phenomenal rate.
Most of your competitors talks about SDS. NetApp, for instance, has its Clustered ONTAP at the centre of their SD strategy. They even claim that they have done SDS year before you developed ViPR.
Srivastava: Scale itself is a very big differentiator. How much does ONTP scale? On a closer look, we can see that it's very limited. The very reason why NetApp is facing this scale issue is, they designed this solution for a scenario that existed many years ago. EMC developed it in 2011 and hence we designed the product for platform 3. NetApp may address the scale part of it, but their design base is such that they cannot address the 'Platform 3' challenge. Their solutions are not designed to handle today's public clouds.
Radhika Nallayam is special correspondent for ComputerWorld India. Send your feedback to [email protected] Follow Radhika on Twitter at @radhikanallayam.