Strategy Automation and Orchestration
Strategy Automation and Orchestration
Strategy Automation and Orchestration
$99
Analytics.InformationWeek.com
S t r a t e g y
S e s s i o n
Presented in conjunction with
CO NTENT S
S t r a t e g y
S e s s i o n
Authors Bio
Executive Summary
11
11
14
Deployment Time
15
The Payoff
16
Related Reports
2 August 2011
CO NTENT S
S t r a t e g y
3 August 2011
S e s s i o n
10
12
14
S t r a t e g y
S e s s i o n
4 August 2011
EXECUTIVE SUMMARY
S t r a t e g y
5 August 2011
S e s s i o n
Automation and orchestration can make IT more flexible and responsive to changing business requirements. Customers use automation tools
to streamline repetitive data center processes such as server operating system and application deployment, decreasing provisioning time for new
services and reducing the risk of administrative mistakes. Orchestration
links a variety of automated tasks together to provision a new service.
The combination of automation and orchestration allows IT deployment to
match the speed of business, providing rapid time to market for new initiatives. But there are also challenges in rolling out automation and orchestration. In this report, we outline successful strategies to automate and
orchestrate IT systems and uncover potential pitfalls to avoid.
In particular, companies must be sure not to underestimate the time or
money required to deploy these products. Also, the complexity of your IT
environment will have a significant effect on deployment and use of these
tools. Finally, you have to prepare the IT staff for the changes that automation and orchestration will have on operations, and ensure theyve been
properly trained on the tools.
We recommend that companies strongly consider the use of professional
services or third-party integrators when deploying automation and orchestration products. Thats because these products require significant integration with existing systems, so having experts on hand can smooth out
deployment bumps.
We also recommend that companies take full advantage of the metering
and chargeback capabilities of these tools, even if you dont bill internal
departments. These options can help IT track resource usage by department, which is critical information when it comes time to allocate budget
and prioritize resources.
Finally, once you reach the deployment stage, notch up one or two easy wins
to help maintain the project momentum. This will also improve the confidence of your IT team to tackle more complex phases of the deployment.
S t r a t e g y
S e s s i o n
38%
6%
14%
We can provision
combined network, storage
and server configurations
with some effort
17%
25%
R2120511/16
6 August 2011
S t r a t e g y
S e s s i o n
enterprise, to order a service. When properly designed, automation and orchestration can
reduce common mistakes, making application or service deployments more reliable through
tested repeatable processes.
The challenge with automation and orchestration tools is that, when improperly implemented, they can cause more harm than good. A mistake in the automation can lead to problems
in the infrastructure. Additionally, automation and orchestration tools are typically complex
and costly to implement, which can stall an automation/orchestration rollout in the enterprise.
Below well discuss common challenges with automation and orchestration, and discuss techniques to surmount them.
Have a goal in mind before starting down the automation or orchestration path. Dont let
yourself get wrapped up in technical detailsstay focused on the final business objectives for
your deployment. End-goal thinking is the best prescription to avoid project-stall.
Figure 2
Automated IT provisioning of
combined network, storage
and server resources
39%
24%
1%
Other
16%
Resource tracking
20%
Dynamic load allocation
R2120511/17
7 August 2011
S t r a t e g y
S e s s i o n
Decrease
5%
Increase
57%
38%
Stay the same
Data: InformationWeek Analytics 2011 IT Automation Survey of 388 business technology professionals,
April 2011
8 August 2011
R2780611/21
S t r a t e g y
S e s s i o n
difficult and expensive the automation and orchestration will be. Additionally, complexity
increases the chances of failure at all layers of the stack.
Our advice: Dont jump into large-scale automation or orchestration projects until your infrastructure is properly designed and ready. Infrastructure is the foundation these tools will be
built on and, as such, is the key to success. Consolidate as much as possible by removing
unneeded components and standardizing on equipment types and vendors. Ideally youll have
as few disparate systems as possibleone storage type, one server type, one switch type and
so on. Remember that every disparate management interface or tool will require another level
of customization for the automation/orchestration platform.
As with any project, the IT department needs to be prepared, trained and ready to adopt the
new processes. The department requires a strong understanding of current manual processes,
scripting, and any automation thats already in place. This should be documented thoroughly
and workflows should be created wherever possible.
Additionally, automation and orchestration will require the organization to adapt, and adopt
new models. Team boundaries will blur or disappear and roles will change. This organizational
upheaval can be one of the most difficult pieces, and must be planned for.
The more mature your operational processes are prior to beginning an automation/orchestration rollout, the better off you are. Having staff familiar with or trained in ITIL or similar
process-oriented systems is valuable, though not necessarily a requirement. If internal staff
doesnt have expertise in automation and orchestration, you may find it particularly worthwhile to consider consulting services to assist in the process.
9 August 2011
S t r a t e g y
S e s s i o n
Once you have the scope in place, set common-sense guidelines around automation and
orchestration. For example, a general guideline might be that the time it takes to automate a
process must provide a return within six months. If a manual task takes one hour of the storage administrators time and one hour of the network administrators time plus one hour for
coordination, you have got a three-hour task. If that task is performed once a week, then in six
months that task requires 72 hours. If it will take 72 or fewer hours to automate the task and
test the results, then you are within your guidelines.
Any task thats not included in the initial scope may always be added at a later time.
Remember that automation and orchestration will provide more time for IT staff to focus on
strategic initiatives once in place. Those extra cycles available after implementation will allow
them to focus on the more complicated processes without bogging down the primary rollout.
Youll also want to decide what you want to deliver from your orchestration catalog. Do you
have developers that need infrastructure, or business units that need database capacity on
Figure 4
Importance of Automation
As data center, storage and application infrastructures get more
complex, how important is automation to the overall future of IT as a discipline?
2011
2009
Critically important
32%
27%
Very important
54%
55%
Somewhat important
13%
16%
Not very important
1%
2%
Base: 388 respondents in April 2011 and 362 in May 2009
Data: InformationWeek Analytics IT Automation Survey of business technology professionals
10 August 2011
R2780611/1
S t r a t e g y
S e s s i o n
demand? These services will need to be designed and integrated into the orchestration layer
and properly presented to users via the portal.
11 August 2011
S t r a t e g y
S e s s i o n
Look for products that can support chargeback. Chargeback is key, even if you dont bill business units for IT services, because it shows the value of the deployment once complete and
provides visibility into who uses which resources. In most organizations IT is a cost center,
and usually has to fight with profit centers for budget approval. If you have accurate, real-time
data on how departments are consuming resources, it will be easier to find sponsors for new
equipment. Additionally, if particular departments use more resources than others, IT has a set
of metrics available for the business to decide how to act (purchase more equipment or scale
back the departments consumption of IT resources).
Where possible, carry out proof of concept. Most any product can look good in a presentation
or demo, but your environment, staff and requirements are unique. Tailor your POC as closely
to your real-world deployment as possible, integrating with or simulating actual workloads
and equipment. With the POC in place, ensure key staff is given appropriate time and
resources to deploy and test within the environment. Its far too easy for IT staff to get
Figure 5
2009
Extensive use
10%
10%
Significant use
30%
33%
Some use
42%
38%
Little use
16%
15%
No use
2%
4%
Base: 388 respondents in April 2011 and 362 in May 2009
Data: InformationWeek Analytics IT Automation Survey of business technology professionals
12 August 2011
R2780611/3
S t r a t e g y
S e s s i o n
wrapped up in the day-to-day and neglect the testing and development time available during a
POC. To avoid this, ensure IT management is on board and understands the importance of
using the POC as a means to ensure deployment success.
Besides the ever present financial considerations, there are several other factors to take into
account when evaluating products:
Integration: Most orchestration software stacks are bundles of other products, new and old,
built in house and acquired. Ensure you understand exactly what your orchestration product
or suite comprises and how the different pieces fit together; chances are, the more separate
products in the suite, the more complexand complexity kills. In addition, examine the way
in which the tools integrate with your existing systems and software, typically API integration
and access points. Your automation and orchestration products are going to have to communicate with a variety of systems in your environment, so avoid proprietary protocols unless they
provide significant enough value to outweigh the inherent lock-in.
Professional Services: Given that automation and orchestration tools are such complex beasts,
youd be wise to set aside budget for professional services, whether from the vendor or a thirdparty consultant or integrator. Even in companies with a rock-solid IT staff, its typically most
cost-effective to have an expert assist in deployment and configuration. Comparison shop, but
dont choke on the price tag: These are high-end services and youre paying to have someone
who knows the ropes do it right, one time, as well as transfer that knowledge to your staff.
IT Skillset: Automation and orchestration tools will require scripting/programming knowledge
as well as an understanding of process definition. Your staff needs to know how to manage the
system, add new services, and build new workflows.
Out-of-the Box Functionality: Custom automation and scripting is a very time-consuming,
potentially budget-breaking process. This extends the ROI period and can tie up IT resources.
Out-of-the-box functionality can alleviate some of this pain by allowing the product to provide
value more quickly while youre rolling out more advanced services. This type of functionality
is typically limited due to the unique requirements of each environment.
That said, tasks such as deploying virtual machines and virtual hosts are typical of out-of-the-box
functionstake advantage of such built-in capabilities. If youre building a private cloud archi-
13 August 2011
S t r a t e g y
S e s s i o n
tecture out of an integrated set of servers, storage and networking, youll find tools with more
out-of-the-box workflows and processes in place because the vendor has done much of the integration work.
Deployment Time
The biggest potential pitfall during the deployment phase is possibility of the project stalling.
If the scope and assessment phase were done properly, youll probably have some easy wins
Figure 6
Automation Challenges
What are the top challenges your organization has faced in automating its data center?
58%
Staff education and retraining
42%
Integrating tools across platforms (servers, networks, storage)
36%
Unforeseen costs in integration/deployment
30%
Lack of suitable automation tools
29%
Platform interoperability
25%
Dealing with change control
21%
Codifying IT processes (i.e., run book creation)
17%
Acclimating management and personnel to automation
17%
Other
3%
Note: Three responses allowed
Data: InformationWeek Analytics 2011 IT Automation Survey of 388 business technology professionals,
April 2011
14 August 2011
R2780611/7
S t r a t e g y
S e s s i o n
as you stand up the automation/orchestration layers and begin applying them, particularly if
you can use out-of-the-box functions. While early wins are key to gathering project momentum, dont use them as a baseline for how the whole project should flow. For example, IT staff
members are typically familiar and comfortable with server virtualization, so automated virtual
server provisioning may happen quickly. By contrast, end-to-end hardware provisioning will
be much more intense and time consuming, so if this phase of your deployment doesnt happen as fast as virtual server provisioning, dont be discouraged.
Use early wins to bolster morale and ensure constant forward push, but be prepared to intermix
some low-hanging-fruit with more difficult projects. This allows you to build confidence and
team knowledge, and the lessons learned in an initial task will be important in deploying more
challenging services. Once you slay a more complicated piece of the beast, it will be impossible
for anyone on your team to bring an it cant be done attitude to the overall project.
Finally, avoid sloppy mistakes during deployment. They will quickly amplify as you automate
and orchestrate. For example, an error in an automated disk-provisioning script is minor on its
own; a single disk isnt deployed or is deployed incorrectly. But when that script becomes part
of an orchestrated workflow, it becomes much more dangerous. If that workflow deploys virtualization hosts from end to end (storage, network, compute, OS and so on) and you test it
after the fact without knowing the automation error, it can take hours to pinpoint and solve
the problem. Test, test and retest at each phase.
The Payoff
The benefits of an automated and orchestrated environment far exceed the risks and costs
when properly executed. They place IT in a position to become a trusted service partner for
business units that need fast and responsive deployment.
Automation and orchestration deployments can be challengingclearly, they are not without
hurdlesbut the tools, resources and knowledge exist and are proven. Know the pain points
going in, prepare the infrastructure, pick the tools and deploy with caution.
15 August 2011
S t r a t e g y
S e s s i o n
Research: State of the Data Center: As data centers face increased resource demands
and flat budgets, we show you how to squeeze more from what you have, and provide
guidance in building a next-gen data center.
Research: State of the IT Service Desk: The primary reason for call-volume increases
continues to be change-related: new hardware, mobile devices, applications, infrastructure architecture and customers. More than 1,200 IT service pros responded to
our InformationWeek Analytics/HDI survey. Dont miss this analysis.
Strategy: Create Your Own Private Cloud: Private clouds enable speedy deployment of IT
services and efficient use of servers, storage and networking. We provide the blueprints.
Strategy: Stack Wars: The fight is on for hegemony in next-gen, omnivirtualized data
centers as vendors including HP, IBM and Cisco erase the boundaries among server,
network and storage suppliers, vertically integrating to create virtual mainframes.
PLUS: Signature reports, such as the InformationWeek Salary Survey, InformationWeek
500 and the annual State of Security report; full issues; and much more.
For more information on our subscription plans, please CLICK HERE
16 August 2011