Virtualization and Cloud executives share their predictions for 2013. Read them in this VMblog.com series exclusive.
Contributed article by Eric Pulier, CEO of ServiceMesh
2013 Predictions: The Application-Centric Enterprise
Companies are under increasing pressure to lower operating
costs while accelerating productivity. One key measure of productivity is
speed-to-market. Another is the cost of managing operations. In a time where
growth is stagnant and margins compressing, a failure to transition to a more
effective model threatens not only earnings, but the future of the organization
itself. To address this problem a cloud-operating model has become the target.
In 2013, every commercial organization and government will drive toward these
goals.
The Age of Cloud IT
Transformation
These Enterprise IT objectives can now be summarized as five
key principles: Transform, Standardize, Automate, Migrate, and Compress.
1. Transform: Transform the IT
organization from a provider of packaged goods to a broker of internal and
external services.
2. Standardize: Drive standardization
wherever possible to decrease management costs; prevent vendor lock-in; and
realize commodity pricing across an ecosystem of service providers.
3. Automate: Automate everything possible.
Eliminate the need for human intervention in the service provisioning
lifecycle, and deliver self-service IT to the business.
4. Migrate: Migrate existing applications
and direct new applications onto standardized platforms. Exceptions are the new
minority.
5. Compress: Compress the Software
Development Lifecycle.
While it is still early in the global transition to this
model, there are now sufficient examples to understand the necessary
ingredients for success. The case studies now available reveal key insights,
strategies and guidelines. The most profound and important of these is also the
simplest: "You don't make money with infrastructure, you make money with
applications."
Infrastructure exists
for one purpose: to serve the needs of applications.
Throughout the world's largest companies, there is a
realization that an infrastructure-centric approach to IT will not scale to
achieve the organization's economic goals. Last-generation IT automation tools
lack an inherent understanding of applications. These tools do little more than
add an unending supply of more workflows to manage-complexity that makes the
problem worse. This common yet disastrous path is known as "Swallowing the
Fly." This strategy states: continuously bolt on more tools to fix the problems
created by the last tools you bolted on, without changing the model itself.
Journey to the
Iceberg
The enterprise journey often starts with virtualization, and
is driven by companies and individuals whose background and incentives are
related to infrastructure optimization. The natural progression in the past was
to go "forward" with a set of infrastructure related goals within the
infrastructure team, with the aid of infrastructure vendors and legacy
infrastructure automation tools. It is often years before the business wakes up
and realizes that it's Swallowing the Fly. While IT continues to chip away at
the problem in the boiler room, the Titanic continues speeding toward the
iceberg.
The primary building block of an infrastructure-led path is
known as a workflow. A workflow is a set of actions that automates steps that
otherwise would be done manually. These steps can include an almost unlimited
range of actions. You can even write policies into these workflows by
specifying who can and cannot access certain systems, or define steps to allow
or prevent other workflows from kicking off. As there is no end to the types of
actions one can trigger with a workflow, vendors peddling such software correctly
claim the ability to do just about anything. And, in fact, they can... until it's
time to scale.
The problem at even moderate scale is that these workflows
simply replicate (and reinforce) much of the inefficient IT operating model
already in place. It's a step backwards, as people are required in many areas--
often many people-- to provision a complete service. One cannot achieve the
end-to-end automation goals of on-demand, self-service environments with these
kinds of workflows. Automating a complete application platform without human
intervention, for instance, could involve the need to manipulate firewall
settings based on the lifecycle stage of an application, or the need to verify quotas for a developer spinning up the platform, or
to match a self-service request with a regulatory policy to determine
approvals.
When a company considers the operational realities of
scaling, it is quickly confronted with a
reminder as to why they are doing this in the first place: to run applications.
If applications are the unit of importance rather than the infrastructure
itself, every workflow exists only in the interest of serving an application in
some stage of its progression toward release. The painful lesson companies face
as their cloud programs mature is how to manage these distributed and
specifically defined workflows for hundreds if not thousands of applications.
Applications are rarely simple topologies with straightforward, static
characteristics. There are almost always variations across each lifecycle stage
of every individual application; variations across the complex ecosystem of
infrastructure that must be automated to eliminate human intervention from
end-to-end; and variations for every change in policy due to shifting security
mandates, regulatory environments, lifecycle optimizations, and a wide set of
governance rules set and revised on a continuous basis.
This complexity of course can be accommodated with
last-generation tools, by simply adding more and more workflows for every conceivable
variation. This infrastructure-centric approach, however, is a Temple of
Doom--destined to collapse under its own weight. As long as one adds more and
more workflows to an already complex environment, the economic and business
goals will remain elusive. To progress, the organization must move away from
optimizing "IT for IT people", and toward delivering "IT-as-a-service" to the
business itself. In 2013, this shift will grow from acknowledgement among
early-adopter CIO's to a widespread strategic imperative.
The Challenge of
Scale
In a simple company, there are simple ways to achieve
IT-as-a-Service. However, in most companies a next generation management model
is needed-one that is easily extensible to allow for centralized control of the
variability inherent in applications. If there is an environment that is even
marginally complex, let alone a multi-national highly regulated one, there must
be a central view of applications and the associated models, policies,
orchestrations and rules. In other words, you have to manage your
infrastructure in an application-centric manner.
Enterprise environments are inherently dynamic in
nature-changing on every level from the procedures of compliance and
strategies, to the context in which these assets are assembled and presented.
The policies that define complex systems need to be not only flexible in nature
but also easily changed and managed at scale. This demands an extensible
"control-tier"-a highly configurable policy function that can operate down into
the infrastructure, among unlimited internal and external systems and
providers, and directly into the software delivery lifecycle--with no loss of
fidelity across any element.
Many speak of policy or governance but fail to accommodate
the most fundamental requirement: the ability to extend the meta-model (i.e.
adding unlimited attributes to application objects that can be set and
referenced by policy) so that a company can not only enforce rules, but make
and manage their own. Traditional vendors approach the matter with a last
generation construct--by allowing customers to set parameters within pre-set
policy types in various places within the product suites, or by writing a new
workflow for every variation. These policies are not only rigid in definition,
and decentralized in management-but they are infrastructure focused, rendering
them essentially useless against the goal of orchestrating applications and
directly empowering developers.
By approaching the problem top down instead of bottoms up, a
new generation of IT management platforms are emerging. A cloud-operating model
demands a single-source-of-truth where one can define, house, manage, monitor
and ultimately orchestrate applications.
The New Cloud
Operating Model
Companies are taking their power back and becoming the
governor of their own service-provider ecosystem. This drives contestability
and standardization into the system and drives friction out. IT organizations will provide services
instead of packaged goods, and their customers will grow accustomed to
consuming SLA's. The result will be widespread margin compression across the
infrastructure vendor ecosystem, as buyers no longer accept value pricing for
commodity goods and services. For the first time, market forces will set the price
for like offerings. Standardized compute SKU's will emerge and high margin
pricing will be paid only for value-added services.
The enterprise has spoken. The 80% or more of IT budgets
spent keeping the lights on will be redirected to new projects and innovation.
Mainframes and other big iron platforms along with heavyweight management and
middleware of the past will persist--but as exceptions not the rule. The target
architecture is cloud-native; and the target platforms managed as standardized
offerings will become the rule not the exception.
The Control Tier:
Gateway to the Enterprise
The "Application Driven Data Center" is being accelerated by
the global economic downturn and enabled by the control and abstraction that
next generation management platforms provide. Every element of the rapidly
standardizing compute stack will need to be centrally managed via policy and
delivered to the business as a utility. This new breed of cloud management
platform will act not only as a control-tier, but also as a gateway into the
transformed enterprise. To survive, vendors will need to transition to provide
services that can be consumed in this model.
While by definition the governance tier must be
vendor-neutral, this does not mean that opportunities to provide services to
the enterprise will be equal. On a regular basis, new on-demand service
offerings will be introduced. Some will be duplicate services that an
enterprise wants to have available to ensure contestability; others will be
targeting specific business or platform functionality. The vast majority,
however, will be distinguished in harder to discern ways among a series of
competitors all making similar claims. For instance, a company may be pitched
on four different "hadoop-as-a-service" offerings, or two different "document
and media management" systems, or three different "API Management-as-a-service"
offerings.
Adopting one service or another is not as easy as "signing
up" like a consumer service. On the contrary, to provision an internal service
an enterprise must locate infrastructure, attain capacity, address security
requirements and allocate human resources to work with the vendor to install,
integrate, run and test a system even for a simple pilot. The planning,
inevitable delays, wrangling among different departments, etc. creates a
lengthy and painful process to even try a vendor's product in-house let alone
buy it. The general direction of services over time will increasingly trend
toward external providers (i.e. Workday, ServiceNow, Salesforce.com), but that
does not mean that all complexity is going away. External SaaS applications
need to be integrated, secured, managed and audited. In addition, many
applications will still be run wholly or partially in-house and delivered to
the business as a service. Vendors will need to crack the code of how to
provide services to this new buyer-one that is brokering services with control
of its own provider ecosystem.
Imagine a vendor and buyer in the future having this
conversation:
Vendor: "If you are interested in this service, we
can get you going with a trial."
Buyer: "Okay,
let's see about getting you into a pilot. I'll look into available resources
once you let us know what your requirements are for infrastructure, operating
systems, storage, database, and access control. We'll set up a meeting for you
with the security group as soon as they can make time. If you are SaaS, I'll
get our identity federation team to begin looking into setting policies for
onboarding a trial and security to review your audit and data protection
policies."
Vendor: "Yes,
you should do that-- for the competitors. To try this service, however, it's
already available in your enterprise App Store. It's set up to utilize
available capacity, pre-integrated with your existing systems, fully compliant
with your policy and governance model. In fact, it's already on your
procurement schedule as an option and is compliant with the same security and
auditing policy that we've established with your existing application control-tier.
You can start using it and if you decide to adopt it, the existing procurement
schedule will allow you to add this additional fee to the existing
subscription. We'll just add something per seat to your existing subscription."
In a world where the majority of scenarios will value
everything below an SLA line on price, success will come from breaking through
the clutter with differentiated higher-level services. Ease of consumption will
be a critical factor for success. Among the existing vendor landscape, the
power will shift to those nimble enough to shift from protecting existing
cash-cows to establishing a channel into the post-transformation enterprise.
Successful vendors will understand and exploit two key
opportunities--an ongoing billing relationship that allows for new offerings to
be added with low friction (analogy: adding a new paid feature to your
Salesforce.com account); and access to the governance gateway to lower the
"time-to-try" vs. competing offerings. These
vendors will populate an organization's service catalog proactively, with
governed orchestrations that can be consumed in a fraction of the time and
price of anyone else. In a world where the buyer needs to move faster to
survive, lowering the friction to sale is a powerful differentiator and the
control-tier will become the gateway to high-margin/high-value services.
In the enterprise, the operating model is changing and
suppliers will follow. The mandate is to transform the metrics of productivity
and cost. For vendors, the race is on to
avoid commoditization, develop new high value services, and to secure a channel
to the next generation customer. In 2013, we will see the emergence of the
application-centric enterprise, and along with it a new dynamic in the rapidly
evolving IT industry.
###
About the Author
Eric
Pulier, Chairman & CEO, ServiceMesh
Mr.
Pulier is recognized as among the leading and most successful entrepreneurs in
government and enterprise technology. The best-known venture capital groups in
the world have financed companies that Mr. Pulier has founded or co-founded.
These include rich media presentation (MediaPlatform), Enterprise Professional
Services (US Interactive), virtual desktops (Desktone), and service oriented
infrastructure (SOA Software). Named one of 30 e-Visionaries by VAR Business,
Mr. Pulier is a popular public speaker at premier technology conferences around
the globe. Pulier is member of Bill Clinton's Clinton Global Initiative, and is
the Executive Director of the Enterprise Leadership Council