What do Virtualization and Cloud executives think about 2011? Find out in this VMblog.com series exclusive.
Contributed Article By Toby Owen, Product Manager for Hybrid Hosting Solutions, Rackspace
Cloud 2011 – The Year of Enterprise Cloud Adoption
Section 1: Enterprise Adoption Begins - FINALLY!
2011 is the year when enterprises finally start truly adopting cloud. That's the big theme for 2011. Before we get to what that means and why, though, let's start off with a working definition of cloud, at least for the purposes of this piece. Cloud means any on-demand virtualization compute platform that can be dynamically sized or resized, and can be provisioned on demand. It includes public cloud, private cloud, on- and off-premise (meaning at a company-owned facility or hosted). For this article, public cloud means cloud at a cloud hosting provider.
"Enterprise adoption" will happen in 2 streams. First is on-premise adoption, which will really accelerate next year, and that is all about the so-called private cloud. This is more than an internal virtualization project though, this is a purpose built multi-tenant cloud, where the tenants represent various entities within the organization. This private cloud allows for resource usage tracking and utility "bill-back" cost recovery, rapid provisioning, dynamic and programmatic access to resource configuration - all the features of a public cloud, but on the company's property, exclusively for use by the company's staff. The second stream is public cloud, where adoption finally starts in earnest. This means production workloads, mission critical apps, starting to move into public cloud usage. The biggest enabler here is hybrid configurations, which are emerging now, to allow physical and virtual compute resources to be connected within a single application architecture.
To frame what "enterprise" means in the context of a cloud marketplace, let's look back at the last 2 years. Virtualization certainly reaches back much farther than that, but cloud computing as we know it today really came into the mainstream sometime in the last few years - at least that's when we saw multiple viable competitive offerings in the market for public cloud. By far, the largest segment of early adopters were small companies, or small groups (developers mainly) within larger companies. These are groups that NEEDED the benefits of cloud - rapid provisioning and cost savings over dedicated hardware - and were willing to accept the shortcomings of this nascent technology to gain those benefits. A majority of cloud users even a year ago were very small companies, where the CEO was also the primary technical user of the cloud instance.
Since inception, CLOUD computing has had some significant barriers to adoption, especially for the enterprise market. Security, performance, all-or-nothing migration, and fear of vendor lock-in are some of the top barriers. It is the solution to these same barriers that will fuel the enterprise migration to cloud in 2011.
Section 2: How enterprise barriers are being addressed in 2010
The first big trend for 2011 that will help to conquer barriers to enterprise cloud adoption is the merging of physical and virtual. This will make some significant steps in 2011, but will take several more years to really mature. Hybrid Hosting models are the primary engine driving this blurring-of-the-lines. Allowing connectivity between on and off-premise dedicated physical server resources and cloud-based compute resources provides more options to fit the right workload or application to the right platform. Developers start to think in terms of which platform works best for which piece of their app, and the added flexibility will spur some really creative architectures which weren't possible even a year ago. This hybrid model allows companies to begin to move their applications to a cloud platform a piece at a time, and avoid that expensive top-to-bottom rewrite, or at least split it into phases. The low hanging fruit which will be first to move to the cloud are batch job processing, the web and storage tiers, and test and development environments. By connecting back to the physical server infrastructure, companies can avoid the more intensive database re-architecture projects and keep their relational DB on fast I/O servers.
Also, private and public clouds begin to blur. As private clouds exhibit more and more of the functionality of public clouds and as public cloud computing adopts more capabilities that were historically relegated to the physical resource world, like load balancing and firewalling, and as tools emerge that can monitor both types of platforms from a single interface, the distinction between private and public clouds start to fade. Further automation of the hardware infrastructure - provisioning, deprovisioning, system management and maintenance tasks - also helps to pare down the list of differences separating dedicated physical and shared virtual compute platforms.
Portability of virtual systems and workloads, which in part is enabled by the emergence of credible open-source cloud projects, is another significant trend which will help enterprises shift more to the cloud. Several open source projects are driving the market toward standards for cloud computing. Some notable ones in the past few years are the Open Cloud Consortium (OCC), which is driving the development of standards and frameworks for cloud, and Eucalyptus, an open source cloud toolset to enable enterprises to build their own cloud infrastructures (a la private cloud style as discussed above). The most notable advancement to date though is OpenStack. This is a consortium of large and small technology vendors and partners who are collaborating to create open source compute and storage platforms. This will enable enterprises to build their own virtual infrastructures privately, or host them in public cloud, without fear that they are going "all-in" with a single vendor, since they'll have the flexibility to move those instances and workloads from off-premise to on, from one provider to another, and so on.
Security and Compliance might be the biggest objection to cloud adoption today, and this is especially true with enterprises. Expect to see some significant headway here. These advancements will be primarily made through software overlay solutions - toolsets that wrap around public and private clouds to provide auditing and reporting capabilities, to provide secure tunneling and encryption management capabilities, and to consolidate systems management of physical and virtual servers into a single console. Service providers will continue to develop their own integrated management platforms, portals, and toolsets, but the niche software development community will provide the best tools in 2011 for addressing these needs. And of course, as I discussed already, the continued progress towards open standards and open source solutions will accelerate the rate at which cloud computing solutions enhance their built-in security capabilities well.
But the crux of the security problem in a virtual world is network isolation. Today, network isolation at layer 2 is essentially the definition of network security, as well as the measuring stick for PCI/DSS compliance. Without it, shared (multi-tenant) hardware won't really approach the security capabilities of physical, dedicated hardware solutions.
Which leads me to the last advancement for 2011: network virtualization. This area of innovation holds the keys to the kingdom for unlocking the true power of a virtualized, utility-based computing future. This will allow on-demand configuration of network conversations, as needed, to be isolated and private, solely accessible to sender and receiver, and then be dynamically reconfigured for the next conversation. While this panacea is the goal, 2011 won't see this achievement, but it does promise some solid progress. Virtualization will definitely move beyond the hypervisor, and begin to flatten the switching tiers in the datacenter. But network virtualization solutions will largely be dependent on a single-vendor solution. These vendor-specific approaches will gain traction, but no standards-based, vendor-agnostic reality will emerge yet. As for API access to the network, we'll start to see pieces emerge, network-service level capabilities such as routing, load balancing and traffic management, security policy provisioning and adaptations based on threat detection. But this will be a disintegrated bag of features for 2011. And most probably, the potential for many companies to adopt these new network innovations will be largely stalled out in 2011 by the need to finally tackle the IPv6 conundrum, since most count-downs for IPv4 address space running out have some date ending in 11. The silver lining in this v6 scramble, though, is that many will be forced to budget and spend on new network infrastructures to be v6 ready, so this will be a chance for them to begin to invest in platforms that will incorporate more and more virtualization, setting us up for faster progress in the years beyond next year.
2011 will definitely be a big growth year for cloud, and with the huge IT budgets of enterprises entering the market in a significant way, the capital at stake only promises to heat up the battle for hearts and minds in the new cloud world.