IBM Global Retail White Paper
Cloud computing for retail
Cloud Computing
2
Cloud computing for retail
Retailers constantly strive to improve efficiency and reduce cost. Managing their large application portfolios has become increasingly complex and expensive and their IT departments frequently must support vertically-focused applications with overlapping functions. Distributed infrastructures and multiple architectures “inherited” from packaged solutions complicate the management of information. As retailers add new capabilities and enter new market segments, the challenges of complexity and cost of managing their IT increase. This paper explains how cloud computing can help retailers address these challenges.
Why cloud computing for the retail industry? Simply put, cloud computing can reduce the IT costs of managing existing and new systems. We see retailers constantly exploring new business models and adding new capabilities to their application portfolios, which in turn increases the complexity of IT infrastructure, volume of data and demands more computing power. For example: •
• •
•
IT challenges in the retail industry
• “Always on” requirements for customer-facing IT services such as websites
• Significant variances in computing capacities between normal and peak workloads
• Flat IT budgets having to absorb new capabilities and growth
• Consolidation of software, hardware and systems management
• A lack of internal expertise
• Efficient reuse of hardware investments (as test and development environments)
• Support for more computationally intense services
•
•
•
Grocery chains are increasing private label assortments and developing new product life cycle management, manufacturing and recipe management capabilities. Food safety requires specialized systems to track inventory. Drug and pharmacy chains, offering medical care and private labels, now behave almost like pharmaceutical companies and require new applications. The use of social networks to increase business requires specialized vertical systems to manage customer reviews, unstructured content, digital media and more for websites. Fashion apparel retailers focus on localized assortments, shorter cycle times and customized offers that require specialized inventory, faster execution of sales promotions and greater visibility of the supply chain. General merchants offer more consumer electronics and subscription plans with specialized application needs. “Do-it-yourself” retailers continue to develop more precise solutions for delivery scheduling, installation and setup, including optimization of field service management.
Such a customer-focused retailing approach creates a tremendous amount of data — driving a 54 percent growth in storage shipments every year.1 Yet retail IT operating budgets, as percentage of revenue, are one of the lowest in the industry. The need to do more with less is therefore a continuous imperative for these IT departments.
IBM Global Business Services
3
At the same time as they are ramping up their capabilities, many retailers are trying to cope with expensive, complicated IT infrastructures. On average, 70 percent of retail IT budgets is spent maintaining current infrastructures.2 Annual operational costs (that is, power, cooling and management) of distributed systems and networking can exceed double their acquisition cost — and all these costs continue to increase.
What is cloud computing?
Even worse, the utilization rates of these commodity servers hover around 5 percent to 15 percent, on average — wasting capacity. This means that as much as 85 percent of retail computing capacity sits idle in distributed environments. It can take months to provision new servers (from purchase to installation), hampering lines-of-business efforts to respond quickly to competitive threats or new opportunities.
Cloud computing focuses on:
To minimize such disintermediation, retailers must reinvent their data centers and infrastructures so that they use virtualization and consolidation to improve server utilization levels with a corresponding reduction in power consumption. They need a data center model that dynamically provisions IT services in minutes or hours rather than months (and at lower cost). Cloud computing is a compelling answer to many retail challenges. For example, recent analysis indicates that, in some cases, private cloud implementations built around larger virtualized servers are almost 90 percent less expensive than a distributed, stand-alone server approach.3 Cloud computing also offers virtualized storage at significant savings and efficient service management and automated provisioning, which can lower the costs of adding IT resources than doing so with traditional mechanisms.
Cloud computing is an emerging computing model that offers users access to their applications from anywhere, with any connected device. The services themselves are in data centers where computational resources can be dynamically provisioned and shared to achieve significant economies of scale.
•
Virtualization of infrastructure and services
•
Automated provisioning of services Elastic scaling (increase or decrease) of computing power—on demand Increased availability and connectivity with end users
•
•
There are different types of cloud offerings — public, private and hybrid—all of which can help retailers better manage their application growth and computing costs.
Public The infrastructure in a public cloud is owned and managed by an organization selling cloud services and is made available to the general public. In this model, computing capabilities and services such as standardized business processes, applications and infrastructure services are accessed by multiple subscribing clients on a flexible, pay-per-use basis.
Private The infrastructure in a private cloud is operated solely for a user organization. The organization can own the private cloud or they can engage a third party to host it—either on site or off. A private cloud provides restricted access to the computing capabilities and resources to be shared by employees; internal departments such as human resources, IT or marketing; and external partners such as distributors and manufacturers.
4
Cloud computing for retail
Private cloud computing helps drive efficiency, standardization and best practices in the services it provides while retaining greater customization and control than public clouds would permit.
Hybrid The infrastructure in a hybrid cloud consists of a combination of both private cloud and public cloud features. In this model, computing capabilities and resources are owned and maintained by both the user organization and the cloud provider. An organization uses public cloud computing capabilities and services for general computing, but stores customer and sensitive data in its private cloud to ensure security.
Choosing your cloud To compare the benefits of public clouds with those of traditional dedicated services, you should evaluate the software services provided by application service providers (ASPs) compared to the software services that are available on the public cloud. Similarly, for private clouds, you should compare the traditional hosted enterprise IT infrastructure available in data centers to services that are available from a private cloud.
Dispelling the misconceptions Not every browser-based application accessible over a network can be considered a cloud solution. Doing so dilutes the value proposition of cloud, which is about flexibility and efficiency of computing resources and cost. However, from a business perspective, cloud, especially a public or a hosted one, can also be considered a different commercial model, one where resources can be purchased by usage, shifting the expenses from a capital expenditure to an operating expenditure bucket. Despite the popular assumption that all clouds are “pay by the drink,” the marketplace is realizing that there is tremendous value in private clouds. Retailers don’t necessarily want to give up control, their competitive advantages or their differentiated capabilities and move to shared, standardized capabilities that puts them on-par with their competition. Retailers would rather simply avail themselves of the benefits of cloud, without giving up control. In such situations, a private cloud is an excellent choice. Not all aspects of the solution need to be “cloud.” Traditional software vendors continue to struggle with how licensing would work in a pay per use model. It is still possible to develop a cost-effective solution where the infrastructure can be on a cloud and the software is licensed by traditional means.
IBM Global Business Services
5
Types of services provided by cloud computing Cloud computing provisions and delivers standardized IT services to users over a network (Internet or intranet) in a flexible pricing and usage model. The users are only aware of the service. They have no need to understand details of the underlying IT infrastructure, its technology or implementation. The service provider is responsible for implementing the service and managing the required infrastructure. In addition to provisioning and virtualization, the cloud model also deprovisions these services so they can be reallocated for other purposes. The concept of repurposing and reuse is a key tenet of cloud computing.
The cloud infrastructure focuses squarely on efficient utilization of the base infrastructure (Figure 1) This includes virtualization, routing and storage management. The cloud platform manages the services running on the infrastructure. The services provided by the cloud are what the consumers actually use.
Business Process as a Service Business Process as a Service (BPaaS) is when a third party leases business processes and capabilities to a company so that the company does not have to handle them in house. The services are available by network and the company either pays as it uses them or makes a low, upfront investment to get started.
CLOUD TYPES
Business Process as a Service (BPAAS)
Business processes as a service; e.g. indirect procurement, payment processing etc...
Software as a Service (SAAS)
Software as a service e.g. email, CRM, eCommerce, merchandise optimization etc...
Platform as a Service (PAAS)
Application servers, databases, middleware, development tools as a service
Infrastructure as a Service (IAAS)
Infrastructure such as servers, storage, file-systems as a service
VIRTUALIZATION
Figure 1: Services in a cloud computing architecture
STANDARDIZATION
AUTOMATION
Reduced Cost
6
Cloud computing for retail
Software as a Service Software as a Service (SaaS) is the distribution of software hosted by a provider in a central and remote location and made available to consumers over a network. SaaS uses a pay-asyou-go pricing model, which decreases or increases the number of software licenses based on need, without having to procure, install or maintain software or hardware or incur ongoing maintenance costs. When retailers use the SaaS delivery model, they can access, business applications, such as accounts payable and customer loyalty virtually.
computing do for a retailer?” Some of the most noticeable benefits that cloud computing can bring to retailers are: •
•
Platform as a Service With Platform as a Service (PaaS), the complete application development and deployment platform (both hardware and software) can be delivered as a service, typically over the Internet. Developers can create, test, deploy and host applications quickly without having to bear the cost and complexity of buying and managing the underlying software and hardware. PaaS is often referred to as “cloudware.” In some cases, Web services, Web 2.0 capabilities and middleware are offered as an integrated platform on which applications can be built, assembled and run.
Infrastructure as a Service Infrastructure as a Service (IaaS) provides hardware components such as servers, network equipment, memory, CPUs and disk space. With IaaS, a retailer could run all operations without installing and maintaining in-house data centers. The approach to the delivery of these services varies from providers to provider.
The benefits of cloud computing for retailers Cloud computing provides dynamically scalable and virtualized resources as a service. Users don’t need knowledge of, expertise in or control over the technology infrastructure that supports them. Whether you understand all the technical details involved or not, the crucial issue is: “What can cloud
•
•
Faster deployment of new capabilities. The use of a common repository, combined with scheduling and automation, means that new capabilities can be deployed much more rapidly. This is due, in part, to faster testing and ensuring tests are thorough and complete (for both unit and integration tests). Improved consistency and quality of new capabilities. A common image repository — one in which common and reusable images are tested and hardened — ensures consistent, higher-quality results. Increased efficiency in the use of IT resources. Cloud computing helps provide significant reuse of existing compute, storage and data resources by simplifying access to them. Faster integration with partners, vendors, customers and suppliers. A cloud-based test environment requires standardization and consistency. This approach allows external partners (such as outsourced development firms) to plan test phases more efficiently and confidently, because the environment is consistent and well known.
How retailers can put cloud computing to work Savvy and market-leading retailers are aggressively pursuing the use of cloud computing, primarily for cost reduction, speed-to-market and quality benefits. The key to doing this is to not “boil the ocean,” that is, create a cloud enablement plan that is so complicated it’s hard to know where to begin. Getting precise about which retail capabilities can benefit from cloud and moving specific workloads will be lot more effective than simply trying to “cloud enable” an entire IT infrastructure all at once. IBM’s methodology, for example, is to take a Component Business Model (Figure 2) view of a retail enterprise and identify specific workloads that match the characteristics that are most suitable for movement to a cloud infrastructure.
IBM Global Business Services
7
Retail functions suited to cloud computing include: •
•
•
•
Merchandise analytical functions that require significantly varying compute power can benefit from server and storage virtualization Store back-office functions running on desktops can benefit from desktop virtualization Card-based payment processing by settlement can be moved to a shared, public cloud for PCI-DSS compliance and better privacy Development and testing of retail applications can benefit from an infrastructure-as-a-service cloud
When trying to compare the characteristics of a service suitable for private clouds versus a typical enterprise data center, it is useful to examine the workload characteristics of the service. If these characteristics can benefit from operating on a virtualized, self-provisioned platform with elastic computing capacity, you can safely and profitably move those workloads to a private cloud. Note: Getting specific workloads to move to cloud is a practical way to get started. The economics of cloud become more attractive when large chunks of workloads are migrated, however, because the virtualization, automation and service management can take advantage of the varying SLA requirements to maximize the usage of the computing resources.
DIRECT
Marketing and Customer Management
Merchandising and Product Management
Strore and Channel
Supply Chain
Business and Finance Administration Corporate LOB Strategy
Customer Relationship Strategy
Brand Strategy
Store and Channel Strategy
Supply Chain Strategy (DC and Distribution)
Consumer Segmentation
Merchandise Strategy
Store and Channel Design
Transportation Planning
Marketing Strategy and Planning
Price/Promotion Strategy
Space Planning
Sourcing Strategy (Logistics)
EXECUTIVE
CONTROL
Supplier Strategy
Financial Management Planning Real Estate Planning Alliance Strategy
Customer Behavior Modeling
Product Development/Design (PLM)
Store Ops Management
Transportation / Fleet Management
Market & Competitor Research
Merchandise Planning
Markdown Management
Supplier Terms Administration
Segmentation Management
LOB Hierarchy and Cluster Management
Planograms
Customer Service Management and Assessment
Price & Promotions Management
Workforce Management
Campaign Management
Category Management
Allocation
Inbound and Outbound Logistics Perpetual Inventory Warehouse Management
Business Performance Reporting Legal and Regulatory Compliance Real Estate Management Stock Ledger HR Management/Career
Customer Service (pre & post sale)
Open to Buy (PO Management)
Stocking and Replenishment
Transportation / Fleet / DC Ops
HR Admin/Payroll
Call Center (multichannel)
Trade Funds Management
Resets and Price Changes
Loss Prevention
Corporate Audit and Accounting
Marketing Execution (Loyalty, Mass, Targeted)
Demand Forecasting
Service Delivery
Returns and Reclamation
IT Systems and Ops
Customer Repository
Master Data Management
POS Execution/ Cash Management
Physical Inventory (Product Track and Trace)
Credit Ops
Customer Communications
Markdown Optimization
Web and Mobile Execution
Supplier Performance
PR Investor and Relations
Labor Scheduling Resource Optimization
Figure 2. Retail component business model
Customer Experience
Smarter Operations
Merchandising and Supply Chain
Other
8
Cloud computing for retail
A private test cloud can reduce costs by reusing existing equipment and limiting future purchases of additional hardware and software. This reuse comes from idle test computing resources and automated reprovisioning and deprovisioning of the hardware and storage as needed for different projects. With the ability to schedule and automate reprovisioning, for example, an existing one-month process for provisioning an operating system (OS) can be reduced significantly. In addition, the result of enhancing the consistency of what actually gets deployed to the testing equipment improves quality. A multi-operating-system image hosting capability and automated provisioning make it possible to maintain a very large number of images and to ensure their consistency with the reduction of manual steps. Exception conditions can still be supported with manual intervention. However, for most of the scenarios, significant error reduction in the deployment of environments to test hardware can be achieved. Cloud computing also creates the possibility of “mashing up” services (such as combining multiple cloud-based applications) to provide higher value-added services for retailers. Consider this benefit when deciding on the best way to begin using the available cloud platform and services. A good way to begin to use cloud computing is to move noncritical services to the cloud first. As confidence improves, some of the more critical services can then become part of the cloud. Some of the other services that are good candidates for the cloud are electronic commerce, test and development and distributed business analytics.
Electronic commerce on a private or hosted cloud An online, electronic commerce channel is a strategic capability for many retailers. In 2009, many retailers experienced flat or lower store sales while their online channel grew substantially. Retailers constantly grapple with cost-efficiently
providing an “always on” electronic commerce channel that can vary significantly in its computing requirements between normal and peak seasons. Electronic commerce, therefore, is a natural fit for cloud because it can provide computing capacity on demand and provision services for ubiquitous access. Furthermore, retailers deal with significant variations in compute capacity between peak and off-peak loads, which are better managed in a cloud environment. Retailers who maintain their own websites and electronic commerce channels invest in the software, hardware and human resources required to manage it. Moving websites and electronic commerce channels to a cloud platform allows retailers to get more out of the underlying hardware, storage and networking infrastructure. If this channel is moved to a dedicated, third-party cloud host, it is no longer necessary to provision for peak capacity and retailers can instead pay for computing capacity based on usage patterns.
Type of
Type of
cloud
service
Private
Information
Benefits
Risks
On-demand
Availability of
computing capacity
services
Centralized IT
Accountability
management of
for loss of
channel
revenue or security breaches
IBM Global Business Services
9
Distributed business analytics
Test and development environments Often in retail environments, test and development environments are periodically underutilized. Maintaining these environments is so expensive that retailers typically prefer to let those machines sit idle until their next period of use. Rapid automated provisioning and deprovisioning of services by moving entire images online and offline quickly are core functions of cloud computing, and this includes entire software configuration environments. These images can also be moved online or offline at a lower cost, allowing the underlying infrastructure to be used for provisioning new services. Type of
Type of
cloud
service
Private
Information
Benefits
Risks
Reuse existing
Cost of switching
investments in
from an existing
hardware and
test and
software
development environment to a setup based on a private cloud platform
Business analytics are a key component of retailing. Metrics help retailers make data-driven decisions about marketing, merchandising and purchasing. Analytics are also increasingly needed for transactions between business partners, such as manufacturers and retailers. When analytics are incorporated into retail operations and supply chain systems, reporting services are more readily available. Retailers need sufficient computing capacity so that additional partners can be added quickly and easily. When a cloud-based platform supports business analytics, retailers can “pay-per-use” or “pay-per-need” for analytics that might be needed less frequently, such as analytics that have higher demand (burst mode) capacity and periodic social network analyses or market analyses. Cloud computing is also an option when a retailer wants a common analytics platform for sharing. information with business partners. Easy access to analytics services on the cloud platform, combined with the ability to adjust computing capacity on demand, makes a multipartner analytics solution a good fit for the cloud. Depending on the metrics to be gathered, data can grow exponentially. Business requirements can often push reporting customization beyond the capabilities of internal development staff. Type of
Type of
Benefits
Risks
Note: IBM also recognizes the value for electronic commerce in a public
cloud
service
cloud. However, that model might not be suitable for all retailers. A public
Private or
Software and Analytics on
Increased data
cloud is a shared infrastructure and a multi-tenant solution. This means
public
information
collaborative
risk because
the retailer’s website will structurally look the same as other retailers on
data is easier to
considerable retail
the same platform. Although the look and feel can be tailored, there are
deliver
data may need to
limits in a multitenant model. Given our experience in working with retail-
be loaded to the
ers who are constantly looking for a highly tailored website that reflects
cloud
their brands, IBM’s approach is to be able to support public and private
Elastic
Up-front data load
models, recognizing that private clouds offer more flexibility that our
computing
effort
retail customers demand.
capacity Lower capital
Difficulty in
expense if using switching providers a public cloud
once data has become entrenched with a particular provider
10
Cloud computing for retail
How retailers can get started with cloud When adopting cloud computing techniques, tooling and processes, it is important to use a phased approach, with each step building on the previous. The initial phase should establish capability, so that evaluation and testing can determine viability, and then enhance those capabilities as appropriate.
Figure 3 shows a roadmap you can use as a starting point. You should then refine it, using detailed information from your organization, such as the specific distribution of mainframe (such as IBM z/OS®), midrange (for example, IBM AIX®), UNIX® (such as Linux® and non-AIX) and Intel® systems.
The initial phase should also address specifically those characteristics of reprovisioning technologies that use automation and virtualization. Begin by identifying and prioritizing cloud initiatives. Evaluate your “cloud level of readiness” to determine what to address first.
Phase 1 Identify Applications
Identify non-Tier 1 Apps for Testing
Define Monitoring Tool Config.
Define Monitoring Metrics
Initial Cloud Service Definition
Protoype Image/Service Repository
Define Provisioning Steps
Define Service Provisioning Policy & Governance
Define SLA Conformance and Governance Model
Define Service Template
Define Usage Accounting Model
Define and Implement Image Management
Define License Management
Define and Implement Update Mechanism
Figure 3. Example of a cloud computing roadmap.
Phase 2 Identify Additional Non-Tier 1 Applications
Storage Address Storage Virtualization
Integrate w/BMC Monitoring Environment
Create Performance Testing Metrics
Establish Backup & Recovery Tooling
Implement Image Management
Update Functional Testing Scripts
Define Security Provisioning Policies
Establish Automation Scripts
Implement License Management
Implement Metering and Accounting Mechanisms
Phase 3
Expand to Tier 1 Application Testing
Establish Performance Testing Env.
Implement and Maintain Security Provisioning Enforcement
IBM Global Business Services
11
Also, include network segmentation and virtualization, along with storage deployment and LUN configurations to further refine this model. The effects on the SDLC, budgeting processes and other issues can be addressed, if appropriate.
Why IBM? IBM leadership in cloud computing extends to delivering enterprise-wide solutions. IBM cloud computing offerings address infrastructure issues with business process services. Three key highlights include:
Conclusion Cloud computing can provide retailers with a list of benefits— today. Retailers should evaluate cloud computing as a viable solution for reducing operating costs, simplifying business processes and collaborating more easily with partners and suppliers.
• • •
Optimized workloads Integrated service management Choice of delivery models
With optimized workloads, workload characteristics determine the rate and degree of standardization of IT and business services. The increasing complexity of IT systems requires advanced service management, visibility, control and automation. Private, public and hybrid models support the wide variety of retailers’ computing environments and system requirements.
For more information To learn more about how cloud computing can help retailers, visit: ibm.com/cloud
© Copyright IBM Corporation 2010 IBM New Orchard Road Armonk, New York 10504 U.S.A
Key contributors Vish Ganapathy is the Worldwide Industry Solutions Architect for the Retail Industry in IBM. He has over 20 years of experience in implementing enterprise business applications in the retail, wholesale and supply-chain industry segments. Vish can be contacted at
[email protected].
Produced in the United States of America June 2010 All Rights Reserved IBM, the IBM logo, ibm.com, AIX and z/OS are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. If these and other IBM trademarked terms are marked on their first occurrence in this information with a trademark symbol (® or TM ), these symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at “Copyright and trademark information” at ibm.com/legal/copytrade.shtml
Martin (Marty) Wolfe is an Executive Client Architect and CTO of the Global Cloud Tiger Team for industries in IBM. He has over 15 years of experience in defining Enterprise Architectures, Implementing Complex Technology Solutions, and defining the link between business goals and IT strategy. Marty can be contact at
[email protected]
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. UNIX is a registered trademark of The Open Group in the United States and other countries. Other product, company or service names may be trademarks or service marks of others. 1
Based on an IBM study for Dynamic Infrastructure/Smarter Planet
2
Based on an IBM study of virtualization technologies on various platforms
3
Also based on the same IBM study of deployed virtualization. Please Recycle
LBW03004-USEN-00