How to Build an Enterprise Ready Cloud
Open Cloud Visionaries: Indu Kodukula of SunGard
This article originally appeared in the January 24, 2012 edition of Forbes magazine, in the
series “Open Cloud Visionaries”.
Indu Kodukula, EVP and CTO
SunGard Availability Services
Open Cloud Visionaries is a series of articles about how IT can navigate the transition to expanded use of cloud resources in a way that meets business requirements, avoids restrictive lock in, and expands choice. In the second article in this series we interviewed Indu Kodukula EVP and CTO of SunGard Availability Services. For an explanation of the context and rationale and a list of other articles, please see: “The Case for Building an Open Cloud” at www.CITOResearch.com
The cloud is part of a long transition for enterprise IT that may not be immediate or revolutionary, but will almost certainly be evolutionary, and significant. The cloud will reverse a trend that has continued for close to 50 years: the continuing escalation of the price of running enterprise IT, according to Indu Kodukula, executive vice president of products and chief technology officer of SunGard Availability Services.
“The biggest economic benefit of the cloud is the utility model,” says Kodukula. “It’s not that the cloud is ‘cheaper.’ It’s that the cloud is more tied to the business value of applications, so that you don’t spend a bunch of money on gear, and you don’t have large capital expenditures on applications that are never going to return business value. Instead, you end up directing your resources toward the applications that are truly relevant and material to the business.”
It’s almost impossible to predict when the transformation of computing — into a consumable resource that is guided first and foremost by business needs — will be complete. It may be anywhere from two to 10 years or more, but there are many reasons for a company like Kodukula’s to be excited about readying CIOs and other leaders of enterprise IT for the cloud now.
SunGard is one of the primary vendors of business-continuity services for large companies whose enterprise applications simply can’t go down, particularly in financial services. SunGard is known for what used to be called “disaster recovery;” in other words, running a phalanx of data centers and infrastructure to support airtight SLAs, which allows for mission-critical applications to run from remote, redundant data centers when the power goes out at the primary site.
But now, companies such as SunGard, Rackspace and others are turning their attention toward supporting enterprise-grade applications in the cloud, which means they’re getting more involved in helping enterprise CTOs optimize their applications–including mission-critical apps like SAP–to run in the cloud all the time. The transition to business continuity has expanded to include protecting applications from run-of-the-mill service interruptions such as system upgrades or errors, denial-of-service attacks or other failures. That is one part of this transition, and supporting this means building a cloud-based system for production.
“We basically made the bet that at some point IT was going to get comfortable to run some applications on the cloud, and that if we built something that was not just for development and testing, but was designed to run production applications, we would be able to get IT comfortable that the cloud is ready for prime time,” says Kodukula. “We needed to build in all the features and attributes that enterprise IT take for granted, in terms of manageability, security, and budgeting, and we made a very deliberate choice to build those in from day one, as opposed to building them on as an afterthought.”
The second part of the transition — and perhaps the hardest one — is convincing CIOs and CTOs, who are paid to be conservative, that the cloud is the way forward for enterprise IT.
Through his many interactions with CIOs and CTOs on this journey, Kodukula has arrived at five major facets of Enterprise Cloud Readiness.
1. Hybrid Infrastructure Management
One of the most important aspects of building an enterprise IT-ready cloud comes from managing infrastructure holistically – which means both inside and outside the cloud, Kodukula says.
For example, SunGard provides Storage Area Network (SAN) services as an add-on to its cloud offering, which most providers don’t. This allows customers to be able to run applications efficiently and scalably in the cloud, but still reliably attach those applications to conventionally hosted databases, such as Oracle.
“This allows you to run your SAP at the level of performance that you’re used to, but pay for that on a utility model, and be confident that your end-to-end SLAs will supported and respected in the cloud environment,” Kodukula says.
2. Relating SLAs to Application Performance
Service Level Agreements (SLAs) are also at the center of providing a credible enterprise cloud service. Importantly, in order to give enterprise IT confidence to run an application in a hybrid or all-cloud configuration, SLAs need to be delivered in terms of application performance, not just hardware or API availability, Kodukula says. SunGard’s SLAs include a recovery time objective (RTO) for applications, which state a specified amount of time for the application to be restored in the event of an interruption, for example.
3. Central Knowledge Repository
To provide effective enterprise IT cloud coverage, vendors must also acquire a deep reservoir of knowledge about not only the cloud, but also about the IT infrastructure, applications and business processes of their clients, Kodukula says. This can involve consulting with CTOs about which applications need lightning-fast RTOs and which don’t.
“I would like to think of us as sort of the Red Adair of IT,” says Kodukula, referring to the oil-well firefighter immortalized by John Wayne in the film Hellfighters. “If you don’t need an application to be highly available, then we’re probably not your first port of call. But if you have a critical operational process that does a billion dollars of transactions a day, then I think we’re in a unique position to offer help, because we understand the best way to design and architect those applications, and how to run them for high availability.”
4. Supporting Applications in the Cloud with Minimal Changes
It’s already a big ask to convince an enterprise CTO to let his application loose in the cloud in the name of getting out of the infrastructure business. It’s an even bigger ask to convince him to make changes to the architecture of that application just so it can run in the cloud. So the best practice is to offer solutions that don’t require such changes, while enlightening CTOs to the possibilities of running more efficiently and effectively in the cloud when, in preparation for a time when they will actually design applications for the cloud.
“We offer a solution that doesn’t require rip and replace,” Kodukula says. “We provide a path for CTOs to take their applications into the cloud in a very measured fashion, and learn as they do that, without letting down their business processes.”
5. Optimize the Cloud for Performance
CTOs fear the cloud for many reasons, including the sometimes erratic performance of the commodity hardware that makes the public cloud such an appealing economic choice. The vendor who’s serious about running an enterprise cloud needs to beef up the cloud for performance. SunGard’s approach has been to run Vblock, the combination package of Cisco servers and switches, EMC storage and VMware’s hypervisor, as the underlying infrastructure of SunGard’s enterprise cloud.
This does come with a price tag. Although potentially cheaper alternatives such as Amazon’s EC2 do exist, running an SAP application in EC2 and trying to connect it to a hosted database through a hypervisor is not likely to be as reliable as when it’s provided by a dedicated hosting service with an insight into the needs of the business, says Kodukula.
These first major steps into the cloud herald a bright new era for enterprises, in which the cost of running enterprise IT starts to decrease, not because the cloud is cheaper, but because it is much easier to tie the correct proportion of computing power to the application that needs it. Over time, a new mindset begins to manifest itself: the enterprise CTO begins to think a little more like a venture capitalist.
The conversation might sound a little like this:
“I want the amount of spend on the application to be $20,000, and the moment I hit that budget, if I’ve not gotten to a level of performance that is acceptable to the business, I want to shut this down, and I want to minimize my losses,” Kodukula says.
The flip side, with allusions to firefighting put aside for a moment, is that “The applications that are taking off like wildfire, I don’t want to limit,” Kodukula says. Instead, based on the same performance metrics, the cloud vendor is pre-authorized to go past a monetary limit, and escalate services rapidly without going through a lengthy procurement process.
“I think this is a pretty fundamental change in how IT thinks about their budget,” Kodukula says.