by Varsha Chidambaram

Sizing up SDx: Indian CIOs Speak Out on the Promise of a Software-defined Future

Mar 12, 201414 mins

The promise of a software-defined future is hard to resist. But getting there will require implementing SDN. There’s where things start to go off script.

Like we didn’t have enough acronyms, the IT industry came up with a few more: SDN, SSDC, SDS or SDT. For the uninitiated that’s software-defined networking, datacenter, storage and transformation. To ease the logjam of letters, Gartner subsumed all of these into one heading: SDx, or software-defined anything.Underneath this alphabet soup is a very real promise: That one day, when every piece of infrastructure in the datacenter is governed by software, all the rigidity associated with IT will disappear.   But to get there, enterprises that have already invested in server and storage virtualization must now implement SDN. Network operators and many technology vendors around the world believe that SDN will herald a new revolution in the datacenter by infusing a layer of automation and programmability that’s never existed before. In their vision, SDN-ville is the last staging post before the wondrous lands of software-defined transformation where flexibility, agility and efficiency bloom.  Let’s hold our stage coaches for a second.   A software-defined tomorrow is great, but without the backing of CIOs, software-defined transformation is a pipe dream. The questions that will define a software-defined future are: Will CIOs invest the time, money and resources to implement SDN and bring about SDx? Is there enough of a business case? Are enterprises unhappy enough with their current networks and datacenters to go the SDN distance?   The Long Road to SDxFor those who believe the idea of software-defined future was concocted in the last year or so, it’s time to burst your bubble. That dream was born when the very first wave of virtualization entered the datacenter.   The advent of the hyper visor was a truly remarkable innovation. It lent underlying hardware new superpowers that allowed it to run and compute over a hundred times the amount it was doing until then.   Crucially, it also tilted the scales of power in favor of software.   Applications now defined and demanded compute resources they needed and hardware was duly provisioned at the click of button (okay, maybe is wasn’t that simple).  But, still, virtualization was an unmistakable gift. Breakthrough applications that offer intelligent analysis of the vast amounts of data we have today couldn’t have been possible without the software-defined concept.   Virtualization also gave CIOs a breather. For long, CIOs were haunted by reports pointing out how datacenter utilization levels were in the 15-20 percent range, and how 90 percent of IT budgets were spent in keeping the lights on. All of this made CIOs look like the caretakers of an inefficient, cost-hungry resource that—while necessary to run the business—didn’t really offer much in the game-changing department.    Virtualization altered that. It kicked off a wave of consolidation that allowed IT departments to pool resources, allocate flexibly, and free up new capacity, which could then be employed to drive ambitious business plans. Suddenly, CIOs had a blowtorch that could melt the rigidity of their steel-cased datacenters, enabling IT departments with newfound levels of responsiveness.   Then came cloud computing, and with it, the bar was raised again. Overnight, business stakeholders—from the CMO to the head of supply-chain—elevated their expectations from IT. What was for the longest time considered a sloppy, error-riddled back office support function, suddenly began to offer what resembled the menu card of a fine dining restaurant, complete with catalogues of services offered, and pairing suggestions, and the cost of each item. IT-as-a-Service became the hottest restaurant in town.  Here’s when things started to get a tad difficult. During the virtualization phase, when doing more with less was the datacenter mantra, CIOs stuffed their servers and storage that they began to suffer from severe bouts of indigestion. They had multiple apps, running amok over multi-vendor tenants, many of which didn’t necessarily speak to each other, and multiple interfaces and dashboards to monitor each subsystem. The cloud magnified this challenge, with its seemingly simple pay-per-use concept. Now, not only did IT have multiple systems and a variety of dashboards within their datacenters, now they also had LOBs buying up more systems in other people’s datacenters.  According to CIO research, 35 percent of Indian CIOs say cloud computing is primarily responsible for increasing datacenter complexity; one-in-four Indian CIOs blame server virtualization. And management complexity due to the use of multiple tools is one of the top three pain points for Indian CIOs with regard to the datacenter.  That complexity is driving CIOs to host and manage their datacenters with outsourcers or cloud providers. A full 40 percent of Indian CIOs say they’re moving less-critical apps to the cloud, and 30 percent are outsourcing datacenter needs completely in order to meet the increasing demands on their DCs.  However, is that the best solution? Perhaps. And yet CIOs can’t fully relinquish control of their datacenters without spending sleepless nights worrying if data was leaking into the hands of the competition. Most enterprises still want to keep their intellectual property within their firewalls. While it’s true that peripheral workloads have moved speedily towards the cloud, a real-life scenario of an enterprise running its entire operation off the cloud is still a utopian idea today.  So what’s a CIO to do? Perhaps the answer lies in an important part of the datacenter puzzle that has been so far overlooked: The network. According to the proponents of SDN, virtualizing the network is the last step towards a software-defined tomorrow, a future in which many of a CIO’s datacenter pains dissipate into thin air.  But for the most part, the network doesn’t get a CIO’s attention. “You need a heart of steel to understand how networking works,” says Sumit D. Chowdhury, president, enterprise ecosystem, Reliance Jio Infocom. Chowdhury is one of the few CIOs in the country who is actively implementing SDN. As a telco that offers 4Gs services, implementing SDN is not a choice but a necessity. If you thought servers and racks were cumbersome, the networking world, with its wires and cables and switches and routers is a whole world of tedium that has, thankfully, remained buried under the floor until now.   SDN—and its inherent SDx promise—is about to change all of that.   SDN: The Last Mile Have you ever played the fiercely-addictive Android game called Flow? For those of you who haven’t, Flow requires a player to establish logical connections between two similarly-colored points, while negotiating traffic amidst several other points, to ascertain the best possible routes for all points, avoiding any conflict or disruption.It’s hard not to wish that the world of networking was as fun as the game. After all, the purpose of all switches and routers is to do just that: To get data to its destination in the smartest way. But if we turned today’s networks into a game, it wouldn’t be fun.   Networking is complicated and drab. In fact, as a subject it rarely gets on a CIO’s radar. Cloud computing, big data, business intelligence and analytics have made demands on the server and storage infrastructure, which, in turn, have re-invented themselves to suit the needs of the business for agility, and scalability. Networking, however, has remained rigid, hardware-controlled and the subject of interest merely among the geekiest.    Software-defined networking promises to change that—and herald in a software defined transformation. But what is SDN?  SDN changes the very fundamentals of how networks operate. For the most part, networks are seen as non-responsive and inflexible. Traditional networks have impaired innovation, while, new-age, high-performance applications are demanding higher service quality from networks.   SDN brings real-time programmability to networks, thus making them more dynamic and better conditioned to deal with fast, frequent changes and increasingly fluid services and applications. Among other things, SDN de-couples the control panel. It separates the forwarding function from the control panel into two different, separately controllable functions. It infuses a layer of programmability in to networking. You can now write code to control functions that were earlier only possible through manual configurations. Finally, it centralizes the networking function and simplifies life. All of this ultimately infuses much needed intelligence into the networking function of the datacenter. This is similar to what virtualization did with servers and storage. Virtualization allowed you to build capacity within existing infrastructure by creating virtual machines, without physically or manually provisioning for more servers. It allowed CIOs to do more with less. The promise of SDN is the same: It removes the tedium and complexity associated with hardware-related functions, orients it to a software-defined approach, and allows CIOs to build agility, scalability and speed into their IT infrastructure. In short, SDN offers a much more user-friendly way to manage network operations without having to actually tinker with the backend.   “It enables people do what they want to do instead of worrying of what happens at the backend,” says Darshan (Dash) Appayanna, CIO, Happiest Minds Technologies.Yet, there are a few hurdles in the way of the commercial adoption of SDN. According to one of the Indian members of the Open Networking Foundation (ONF is a not-for-profit whose mission is to propagate the adoption of OpenFLow), there is a long lead time between what happens in networking labs and solutions sold in the market. To make matters worse, implementations are assumed to be complex, which further slows down SDN adoption. Then, there’s the skill set challenge. Most network operators have been trained to work with a CLI-based approach—not with programmable interfaces. SDN StallAppayana of Happiest Minds is currently in the process of testing SDN-ready products. The company operates a very lean, fully virtualized IT infrastructure, with the bulk of its apps running off the public cloud. But it hasn’t yet gone down the SDN road.   “We haven’t done network virtualization yet, because the technology is neither ready nor is it cost-effective. Like hybrid cars, SDN is still not commercially viable,” he says. That said, Appayana says that SDN is the way forward, “We operate in SMAC (social, mobile, analytics, cloud) mode, and SDN will help us react to our dynamic business needs better. But, it is at least four years away from being commercially viable.”  WATCH VIDEO And even when viable solutions hit the market, it isn’t likely CIOs will rip out their existing infrastructure before the normal refresh cycle. “How often do you change a router or a switch? Unless it fails, you don’t,” says Appayanna. “Network devices usually have a much longer shelf life than servers and desktops. You will change a router or switch perhaps once in 10 years.”   James Berry, CIO, Standard Chartered Bank India and South Asia, says they are adopting aspects of what is now being called SDN. “But, we’re working at a pace that suits us, not suppliers,” he says. Unlike Appayana, Berry is more skeptical of the promise of SDN. “In theory, SDN allows you to leverage your infrastructure more effectively, with the opportunity of engaging best practice and third-party technology. But it could also add another layer of complexity.  A key decision is how you balance SDN with fixing legacy,” says Berry.Chowdhury of Reliance agrees. “SDN doesn’t really remove complexity; networking still continues to be wired. It just removes the complexity from one level of users and makes orchestration and abstraction possible.”  Chowdhury also believes that the market for SDN is limited. “Other than telcos, I don’t think anybody will be doing SDN,” he says. Berry points to the security threats of SDN. “Another important point is that if we implemented an SDN-based solution, we would need to make sure the approach and architecture is robust enough to secure our client data without creating additional risk, and that it is easy to troubleshoot and fix issues.  Ensuring the business case works with those key challenges is very difficult.”   “Data security concerns, both real and unfounded, need to be managed.  The biggest inhibitors for us are the legacy investments we have already made in the infrastructure or services in focus, and banking regulators’ comfort level with it,” says Berry. Both Appayana and Chowdhury agree that SDN is still at least three years away from commercial adoption. “And even then it may never make sense for a manufacturing or a retail outlet,” says Appayana.   So what’s the real business case of SDN? “From my current understanding of the subject, and from speaking with my peers (rather than technology vendors), one thing that is not all that clear is the real benefits case,” says Berry. “Yes, there’s lots of marketing material, and ‘marko-tectures’ that get thrown at you to show you how beneficial SDN can be to your organization. While they make sense, theoretically, the reality is that holistic benefits are much more difficult to achieve than the promises made in some of the literature. You need to carefully weigh the financial and resource investments versus the business benefits when considering SDN,” says Berry.   Proponents of SDN believe there’s a business case for it. Finding out whether your business needs SDN is all about asking the right questions: How dynamic is your business environment? Do you have projects that need to be kickstarted and go live in two days? Are you on a high-growth phase where new offices and employees are recruited frequently? Are you running geographically-dispersed operations that require workloads to be shifted dynamically, without compromising user experience? Do you want to customize and define identity and policy-centric applications to behave in a certain way for a certain user for a certain period? (For more use cases read When SDN Makes Sense)  If your business falls in any of these categories, SDN could offer your organization the nimbleness it needs to spring into action and respond at lightning fast speed. SDN can reduce networking provision time from weeks to hours. It can dynamically shift loads between clouds, often called cloud bursting, to offer the most optimum user experience to your geographically dispersed user base. It can help optimize your resource utilization and do real justice to your already sunk-in investments in virtualization and the cloud.   The Last Word The fact is, the needs of the business are fast surpassing the capacity of IT to deliver it. The only way to keep pace with business needs to re-orient IT to a software-centric model, a model where hardware is controlled and aligned to the application it serves.   Today’ datacenters exist as a ‘patchwork quilt.’ CIOs have chosen to use a more organic approach to datacenters, replacing structures bit by bit instead of ripping and replacing the whole shebang. Incremental changes to the datacenter is sure to minimize disruption. But it can never create the transformational business value or competitive differentiation of a big bang approach. And not many enterprises possess the wherewithal do implement such a change. And hence, as with any other innovation or technology, SDN, too, will probably run a gradual course of a slow adoption, accelerated by market offering and business needs over the next decade.   “The service providers are a little too bullish on client acceptance, and utilization.  And given that investment budgets are tightening, rather than loosening, I would see the eventual timeframe elongating even further,” says Berry.   For Appayana, too, it is a wait-and-watch period. “We’re starting to evaluate how mature the technology is today and to what extent it can deliver what it promises.”Chowdhury believes that SDN will not see the speedy uptake that server or storage virtualization saw in the enterprise. “The network evolution story is going to be a long one.”   ALSO READ: Five Reasons Why You Should Invest in SDN Varsha Chidambaram is principal correspondent. Send feedback on this feature to Follow her @Varsaw