By Charles King, Pund-IT, Inc. May 21, 2014
Of the issues roiling modern day enterprise IT—big data, mobility, BYOD, security—few promise as many fundamental changes as cloud computing: including the decision of what and/or when to offload processes and data to a public cloud vs. developing an on-premises enterprise private cloud. That’s partly because while the technological challenges of cloud are significant, it also requires companies to essentially reassess their IT assets, platforms, infrastructures, staff and goals.
Doing so successfully can deliver notable benefits and efficiencies, but the process is anything but easy. Plus, results can vary widely as organizations are pursuing the three integrated cloud deployment models: private (on-premises), public (off-premises) and hybrid (mixing public and private). That said, despite the considerable hype surrounding it, the strategies, technologies, and business models of cloud computing are clearly here to stay.
Public cloud players, including Amazon, Microsoft and Google are certainly helping drive momentum, but the impact of cloud on IT sales is being felt much more broadly. For example, late in 2013 Diane Bryant, SVP and GM of Intel’s Data Center Group, noted that while the company had originally forecast a bullish 37% compound annual growth rate (CAGR) for 2011-2016 cloud revenues, the sector’s CAGR from 2011-13 was actually 47%. Plus, along with delivering significantly better results than other data center sectors (enterprise, HPC, telco), cloud sales are showing no signs of slowing down.
But that success begs the question of whether enterprise private clouds are as simple and successful as they could be. Because the promise of an on-premises enterprise cloud is typically less painful than its actual implementation, what would it take to change that? Just how hard would it be to create what amounts to an “easy” button for developing, deploying and delivering enterprise private cloud?
Why single out private cloud for attention? Mainly because while public clouds like Amazon AWS, Google Cloud and Microsoft Azure are grabbing headlines, the vast majority of enterprise customers regard public cloud as just one piece (and often a very small one) of a much larger picture. In part, the issue is cultural. Many if not most public clouds built their reputations by serving “shadow” IT—services aimed at individuals, work groups and departments that felt the need (justified or not) to sidestep conventional IT channels. Many organizations may tolerate that approach, but only some will embrace it wholeheartedly and few, if any, will consider it for supporting business-critical applications and sensitive company data.
What’s involved in building a private cloud or adapting an existing internal IT infrastructure for delivering cloud-based services? First, remember that cloud is, after all, simply an IT management approach that typically emphasizes virtualization and automation. By that measure, enterprise IT vendors and customers began using cloud mechanisms decades ago in mainframe and then Unix and other enterprise systems. What initially pushed contemporary cloud concepts to the fore was the huge success and uptake of x86-based virtualization pioneered by VMware, then followed by competing solutions, including Microsoft’s Hyper-V and Citrix’s Xen Server.
Following the dot.com bust, Web-based companies also began looking at ways to maximize the energy and cost efficiencies of their often massive x86-based IT infrastructures. Amazon’s AWS, in fact, arose from a 2003 report by company engineers describing a standardized and automated company IT infrastructure that they suggested might also support commercial service offerings. The first public AWS offering—Simple Queue Service— was launched in late 2004. In the years following, rumors and media stories swirled around similar efforts by Google, Yahoo, Facebook and others, and were followed by a growing number of commercial cloud-based services and players.
So if cloud concepts are so well established and understood, what’s keeping enterprises from getting their on-premises private clouds up and running in short order? Three issues are of particular concern:
- Surety—Given the critical roles that computing solutions fulfill in organizations of every sort, contemplating (let alone instituting) fundamental changes requires that core processes and information, including legacy applications and data, will be not be disrupted or destabilized, and that preserving business continuity is a core value.
- Security—The security of critical business workloads and data is one of the primary reasons organizations resist engaging public clouds, particularly when such engagements involve controlling sensitive data, protecting intellectual property (IP) and preserving privacy. So no one should be surprised when IT insists that on-premises private cloud environments strictly follow security, governance and compliance policies.
- Standardization—A key lesson of public cloud concerns the value of standardized solutions and architectures, especially in enabling resource pooling and business agility. Enterprise IT needs to embrace standardized technologies and platforms to achieve the full capabilities of private cloud. But that exercise can also cast light on the value of and burdens imposed by proprietary legacy systems.
If these three points can be successfully addressed, IT still faces critical technological and process challenges related to orchestrating and enhancing asset utilization. Chief among them are:
- Building and managing an enterprise grade private cloud.
- Providing their customers modern platforms for cloud application development.
- Automating deployment, monitoring and run-time management of applications in their cloud.
- The ability to gain cloud efficiencies for their significant base of legacy applications.
Many companies have the in-house skills and expertise required to successfully tackle private cloud planning and deployments. As we noted before, the technologies and processes that support private cloud infrastructures and processes have been part of enterprise IT for decades. Plus, major vendors, including VMware and Microsoft and those leveraging OpenStack technologies offer numerous solutions designed to facilitate private cloud efforts.
But for those who cannot or would prefer not to embark on a self-guided journey to private cloud, dedicated cloud consultants and professionals can provide advisory and deployment services to help ease the way. Here are four companies that specialize in a variety of cloud services and solutions:
- ElasticBox aims to make it as easy as possible to develop, deploy and manage applications for any cloud infrastructure, including on-premises enterprise clouds. The company offers a simplified, modular application development process, where fully-configured architecture components, including scripts, variables, files and other metadata are encapsulated as “boxes” and made available as a service. Since “boxed” applications are decoupled from underlying cloud infrastructures, they can be moved between private and public clouds (including Amazon Web Services, Google Cloud, HP Cloud and VMware’s vCloud) or from one public cloud to another. For enterprises, ElasticBox supports continuous integration and simplifies component updates and patching by pushing changes to all the applications using a specified box
- GigaSpaces Cloudify is a free, open source software platform designed to move existing apps to enterprise clouds with no code changes. The critical component is Cloudify’s ‘recipe,’ a deployment and management plan describing the various dependencies between application tiers, plus the lifecycle events (install, start, configure, stop, uninstall, etc.) and the cloud resources used by each tier. The recipe also determines how to monitor, upgrade, maintain and scale application tiers. In addition, Cloudify’s Orchestrator uses a recipe to install an application on the cloud, from VM provisioning to application code deployment and dynamic wiring. Afterward, Cloudify monitors applications based on customer-defined metrics, and auto-heals crashed middleware or lost VMs while automatically reconfiguring the app. Cloudify also supports post-deployment upgrades and fine-tuning, as well as application auto-scaling capabilities.
- Intigua pursues a simple goal: deliver a production-ready VM in 5 minutes or less configured with its entire critical management stack, including server monitoring, configuration and compliance management, security logging, data backup, etc. The company achieves that with an automation platform supporting centralized, policy-based provisioning, configuration and ongoing management of the entire management stack. That includes technologies critical for SLAs, security and operational best practices such as server monitoring, configuration & compliance management, security logging and data backup. Intigua works with existing management technologies, including BMC, CA, EMC, HP, Hyperic, IBM, Microsoft, NetBackup, Puppet, Splunk and Symantec. Along with simplifying management stack processes today, Intigua also prepares customers to support the scalable, policy-based automation required for software-defined data center (SDDC) environments and standard cloud management platforms, like VMware’s vCAC & vCO, CloudStack and OpenStack.
- Redapt is a full service systems integrator whose data center infrastructure and cloud solutions are custom designed for specific clients. For enterprises, Redapt provides private and hybrid cloud solutions that deliver all the benefits of a public cloud while enhancing overall performance, security, and savings. The company’s cloud experts collaborate with customers to architect solutions tailored to specific compute, storage, and networking requirements, and to eliminate the bottlenecks and excessive operating costs often experienced in public clouds. Redapt’s advanced application modernization and migration services can help businesses transition cloud-ready applications and transform legacy investments into optimized, cloud-savvy solutions.
Crafting Cloud from the Ground-Up
Many vendors are focusing considerable investments and talent on private cloud solutions, but for our money, Intel’s effort is one of the most intriguing and impactful. The company’s strategy springs from its experiences in developing its own enterprise private cloud starting in 2010. Today, Intel’s IT infrastructure spans 75,000 servers in 69 data centers worldwide and supports over 90,000 company employees and more than 138,000 devices, the majority of which are business PCs/notebooks and handheld endpoints. Intel’s datacenter also supports thousands of workloads deployed in a private cloud virtual environment.
The development of Intel’s private cloud followed these steps:
- Our Pain/Your Gain: Intel’s is a classic example of a vendor practicing what it preaches. Early on, the company recognized that private cloud offered key IT advantages in terms of speed, agility and efficiency, while also paving the way for the evolution of data center-enabled services and extensions to external services/providers. Over the course of creating and deploying its private cloud infrastructure, Intel developed a host of best practices and some commercial solutions, including Virtualization Best Practices, Cloud Builder Reference Architectures, Open Data Center Alliance (ODCA) Cloud Usage Models, IT Workflow Automation, Intelligent Power Technology and Cloud Security Technologies.
- Open Source/Open Stack: Intel succeeded in developing one of the industry’s preeminent cloud-ready processor platforms (its Xeon E5-2600 product family). But the company also realized that open source solutions, particularly OpenStack, offered key advantages for cloud, including interoperability across multiple vendors’ systems, speeding development via shared collaboration and allowing partners to focus more on service development than core compute issues.
- Server/System Telemetry: To maximize capital investments and operational efficiency, data center owners need to optimize power and cooling distribution, especially in cloud infrastructures leveraging dynamic workloads. Intel created its Node Manager technology (available first in its “Haswell” Xeon processors) to help users optimize CPU power utilization according to workload requirements, thus increasing compute density while maintaining fixed rack power budgets. Intel Node Manager also supports continuous thermal monitoring and analysis of data center assets, including individual servers, full racks and entire rooms. That helps to ensure that Intel-based systems are performing properly and in accordance with broader cooling, workload and budget priorities. Finally Node Manager can be integrated with Intel’s Data Center Manager solution to deliver critical insights across enterprise cloud infrastructures.
- Developing for Developers: Intel recognized that the end users who are often most directly affected by cloud are enterprise developers. So it isn’t surprising that the company is placing special emphasis on the tools developers need to create cloud-aware applications that take full advantage of cloud characteristics like elasticity, resiliency, run-anywhere design, multi-tenancy and self-service provisioning. Creating cloud-aware applications also requires developers to think in new ways, an issue Intel is addressing with key lessons in developing consumable web services.
Intel’s systemic approach to private cloud is no surprise, given the notable leadership position of its silicon in enterprise servers and systems. But as Intel continues to expand its capabilities in storage and networking solutions, it will be capable of supporting levels of cross-data center component transparency that should make future cloud solutions and strategies ever more innovative, insightful and influential.
While the technological underpinnings of cloud computing have been available for years, the evolution of standardized IT solutions has helped cloud to benefit far more organizations and individuals than ever before. But at the same time, concerns about public cloud security and management methodologies and the inherent complexity of deployment are inhibiting the uptake of cloud by many of the organizations that would benefit from it the most.
We believe those issues could be addressed and adoption accelerated by the development of what amounts to an “easy button” for creating and deploying on-premises private cloud infrastructures. The tools and technologies for instituting such solutions certainly exist, and are bolstered by cloud services like those offered by ElasticBox, GigaSpaces, Cloudify, Intigua and Redapt. Plus, the continuing evolution of core foundational microprocessors and related technologies is central to the current and future success of enterprise private clouds.
That’s where efforts of Intel and its OEM/ODM clients come in. In essence, the company is developing the tools and technologies that server manufacturers and systems vendors need to create innovative standardized cloud solutions. But Intel is also extending its efforts beyond those constituencies to create cloud-related best practices and guidelines that benefit cloud-minded enterprise customers and developers and open source communities.
At the end of the day, cloud will only be truly “easy” when solutions address the requirements and concerns of virtually anyone whose finger presses the button. Successfully meeting the needs of those individuals and organizations is central to Intel’s private cloud strategy and efforts.
© 2014 Pund-IT, Inc. All rights reserved.