Taking a Shot at the Moon- and other cosmic questions

MUMBAI, INDIA: Not everyone can see ‘the' face in a moon, but then not everyone imagines rockets and other dream-machines to reach this faraway bundle of fantasies. In every era, there have been people (poets, scientists, astrologers, astronauts - the race doesn't matter) who can see possibilities and in every era they have been flanked by naysayers.

So yes, again there are companies like HP who are apparently trying to figure new leaps in erstwhile black holes like poor-server utilization, energy and power part of Catch 22, space hogging boxes, and the stubborn gravitation forces of heavy costs when it comes to servers. And again there are ‘buts and ifs' breed of gate-crashers like us, who can't help wondering if the party has started too early here.

While pilots like Stephen Bovis, vice president and general manager, HP Servers, Asia Pacific and Japan, are excited and busy catapulting HP's innovation projects into new heights, we get chances like the HP World Tour to corner them with the baggage of curiosity for a few minutes.

We try our best to hint at eclipsing forces like server-utilization race with virtualization alternatives, rivalry between fuels with dedupe vs. compression, commoditisation of servers, the inveterate cost and power juggle with apples of efficiency and more. Bovis to his credit and with extreme navigation skills coming very handy, fights the shadows of cynicism with the grit and weightlessness of a true-blue astronaut. Despite all that fuzziness we try to whip up, he manages to show us a floating glimpse of the face he sees at HP. Here's a quick Moon-shot:

Has Moonshot been navigating all the anticipation well?

Moonshot has shaped up well to the question of specific workloads and different needs of different customers. Not every customer needs a certain type of server architecture but they all face issues of space, efficiency or power consumption, no matter where. If you can solve some of these big issues, you can crack the problems. We have been trying to offer the right answers and the same force has guided our innovation also.

One of the strongly advocated advantages of Moonshot servers is their effect on server utilisations. How does that compare with what virtualisation vendors purport on the same factor - ability of VMs to increase utilisation rates significantly?

With data center demands increasing exponentially, the server market is relying on game-changing technologies to dramatically improve server performance, management and energy consumption. These breakthroughs require dedicated knowledge, experience and resources that only HP can bring to the table. We believe that Moonshot offers customers a new class of server with a superior value proposition. HP Moonshot delivers a significant improvement in energy, space, cost and simplicity, unmatched by any other vendor.

Does that make virtualization a complementary element or a competitive force here?

We offer virtualization in our solutions and they are popular choices. Virtualisation came in because of the ‘efficiency' question and the struggle with server utilization. Moonshot handles the same outcome but it co-exists with virtualisation. We see IT as converged IT and specific workloads in some verticals are going to drive new adoption patterns.

What would you say is a stronger proposition for this new breed of servers - energy efficiency or customisation or workloads or cost factor? Specially for an Indian prospect?

The explosive growth of social, mobile, cloud and big data in India will be restricted by the very thing that enables it - Infrastructure. HP Moonshot system is a family of software defined servers that are purposefully built to address the data center challenges of social, mobile, cloud and big data. Each server will be designed to address the specific needs for each workload. The HP Moonshot System, with its ultra-small server form factor, uses up to 89 per cent less energy because it leverages low-energy SoC's (system on a chip) designed originally from smartphones versus traditional microprocessors.

Is that Economics redefined?

These processers have lower wattage requirements and a dramatically reduced energy and space footprint. For specific scale-out workloads such as web services, dedicated hosting, content delivery, etc., this model scales workload performance across a large number of servers, with dramatic gains of TCO, and 77 per cent less cost for dedicated hosting.

Moonshot offers 10 times the volume of traditional servers. Ultimately, HP Moonshot System changes the infrastructure economics and overcomes the space, power and cost demands of traditional technology. All of which are important to our India customers.

What's your view of attempts like OpenCompute etc and do you ever see Indian enterprises going the DIY (Do-It-Yourself with OEM servers) way with servers and commodity infrastructures?

Our commitment to some key Open initiatives is steady. As to India, scale would be an important parameter and server provisioning part will matter. It would be about people who can scale in a big way. Still, HP is proud of its leadership position in India and customers admire what we bring, which is a good mix of TCO as well innovation needs. So we are confident and excited, nonetheless.

Any take on Flash's penetration? Like for-point use vs. mainstream uses? How do you reckon questions around Flash's attraction vs. disk specially on areas like scalability, cost advantage and also on dedupe's performance impact vs. efficiency debate?

We're bringing flash storage to the price point of spinning disk, enabling customers to use it for more than niche applications. HP 3PAR StoreServ delivers low-cost flash, performance and resiliency at enterprise scale. 3PAR StoreServ 7450 eliminates flash silos and lowers costs with Thin Deduplication and high-capacity cMLC SSDs. With patented Adaptive Sparing, 3PAR StoreServ reduces wasted space per drive and extends usable media lifespan without sacrificing best in class drive endurance.

What do you make of the dedupe vs. compression debate?

Compression is a capability that serves you well in application data sets that are less dedupe-friendly, like databases. Well as it turns out, zero-block deduplication is also an excellent technology for those database workloads as well. It gives you often a 2:1 efficiency advantage very similar to a compression technology, so we don't need compression to be as efficient as what other vendors might claim in those environments.