Clouds on the Horizon - Part II

Cloud_Computing_1

 

 

 

 

 

 

 

 

 

 

Posted by Chris Iles and Benoit Wiest

This article is the second part of a two-part blog post that discusses cloud computing in the public sphere, especially in developing countries. It outlines the opportunities and obstacles, and discusses key issues to be considered.

Opportunities and obstacles

The proliferation of applications and their underlying server, storage and communications environment places a considerable burden on the IT staff tasked with configuring, deploying, maintaining and managing these assets. Once an IT system is up and running, ensuring its reliability can also be a challenge in developing countries where IT infrastructure may be limited and can be affected by adverse weather conditions.

Cloud computing addresses these issues by simplifying the IT environment and divesting IT organizations of much of the responsibility associated with managing applications, while at the same time increasing flexibility and reducing the total cost of running the applications.

Public clouds lie at one end of the cloud continuum, so they provide a good point of departure for considering costs and benefits, as outlined below. Not all opportunities or obstacles that relate to public clouds necessarily apply equally to other delivery modes.

Public clouds have the following advantages:

Scalability – applications and data can be delivered to any number of users, as needed. This characteristic is relevant to developing countries where the number of functions and transactions needed may be quite low.

Flexibility – providing online access to large data files over any internet-enabled device, avoiding limitations of older workstation technologies.

Fast deployment – new applications can be up and running very quickly.

Reliability, fault-tolerance – large scale cloud server installations provide redundancy (duplicate components cover the failure of individual parts) and ensure availability.

Accessibility – the internet enables coverage of wide geographic areas, establishing the government’s presence across its territory.

Best of breed computing – users have access to automatic software upgrades that may not be feasible to own in-house.

Shared overheads – the burden of managing data centers and supporting end users is placed in the hands of skilled professionals.

Focus on core business competency –freed from data center administration, cloud users can invest in establishing functional excellence in business competencies instead of competing to acquire IT skills in the local labor market.

Reduced Total Cost of Ownership (TCO) –users pay for use, not for the acquisition and support of hardware, software, facilities, and communication networks. (This benefit is less applicable to private clouds).

Simplified expenditure – pay-for-use replaces multi-faceted costs of investment, license renewals, maintenance and user support with single, and flexible, predictable monthly O&M charges. This simplifies budget forecasting.

Economies of scale – large providers can negotiate procurement contracts on preferential terms that may not be available to individual IT organizations.

Security –cloud providers implement multiple layers of protection over physical facilities, applications and customer data that is often superior to in-house data centers which struggle on a daily basis with viruses and other threats to performance and security.

Disadvantages of public clouds:

Loss of control – especially over data access and protection, identity management, system operations, data transmission and flow controls, business continuity and regulatory compliance. (This is less of a concern for private clouds).

Loss of data sovereignty – the customer may not know where data is physically located to ensure compliance with regulations which restrict where government data can be held (Private clouds address this concern).

Vendor lock-in – migrating solutions to a new cloud provider may be problematic.

Dependence on provider’s service level – the customer cannot directly control the provider’s quality of service.

Infrastructure constraints – limitations on telecommunication and power can reduce the effectiveness of cloud solutions.

New costs – because cloud computing is enabled by telecommunications, the associated communication cost can be high.

Challenge to ICT management - by offering an alternative to in-house IT solutions, the cloud can allow business units to bypass official IT standards leading to a disorganized IT environment operating outside the official IT strategy. (Private clouds keep service delivery under corporate control to avoid this problem).

Issues and remediation

As cloud computing is still a relatively new technology, best practices have not yet emerged. Meantime, issues to consider include:

1. Performance

Slow response times (“latency”) can jeopardize user acceptance but can be mitigated by:

(a) minimizing the physical distance between the network node and the end user;

(b) deploying appropriately powerful servers;

(c) optimizing browser interfaces to simplify administrative ‘handshakes’; and

(d) verifying the provider has tools such as metadata catalogs, and processes such as frequent caching and volume-based structured storage, to optimize legacy applications.

2. Provider service level

Careful selection of cloud providers and negotiation of appropriate SLAs can mitigate concerns about service levels.

Providers’ practices should be validated with respect to multi-layer security and controls, hiring and supervision, internal auditing, performance management, data storage and protection, problem notification, business continuity and disaster recovery. References and industry certifications (such as the US SSAE 16 or international ISAE 34020) should be checked, and performance tests conducted to verify actual capabilities against claims.

Negotiating and enforcing a tight Service Level Agreement (SLA) with the vendor is critical. Good legal advice is essential here. Because of the primacy of the users’ experience to a successful cloud deployment, traditional IT-oriented SLA performance metrics should be replaced with measures that reflect the quality of the user experience.

3. Data

Legislation or policy in some countries may require governments to maintain their data within national boundaries. Sometimes the solution is a dedicated in-country data center, as in Moldova and Slovenia, for example. This is not always required. For example, the UK government was reported in 2012 to be opening cloud services to overseas providers for certain data. Organizations whose data is hosted in the EU can get some comfort from the EU Directive on Data Protection and in the US from compliance with the EU-US Safe Harbor Privacy Principles.

Private clouds address these concerns at the cost of losing some of the benefits of the public cloud, in particular sharing overhead, replacing capital expenditure with metered usage costs and economies from large scale provisioning.

4. Costs

The transition to cloud computing should be justified by a cost benefit analysis that compares all expected costs (including the cost of migrating to the cloud and operating costs such as communications charges) against expected benefits. After comparing the cost of different delivery models, public cloud may not be cheaper than private cloud where applications are in constant use.

Vendors such as SAP and Oracle have programs to enable existing customers to transfer their annual support fees towards subscriptions for SaaS1. This is to encourage customer loyalty, but nevertheless may provide a cost-effective transition path.

Once operational, use of cloud resources must be controlled to avoid runaway costs.

5. Governance

There should be an overall policy and strategy to manage the deployment of cloud-based resources. Without a formal goal and a plan to achieve it there is a risk of replacing a complex, in-house data center with a complex, disjointed, hard to manage cloud solution. It should cover the transition, optimization of business processes and job descriptions and address performance, security, control and availability. Plans should take into account infrastructure limitations, notably available bandwidth.

Outsourcing the ICT environment does not diminish in-house accountability. Many parties may be involved in delivering applications, requiring an enforceable governance structure to determine who is responsible when things go wrong and what remedial performance or compensation is expected. Issues to consider include failover procedures, continuity, disaster recovery and how to provide an “off-ramp” to a subsequent new platform.

6. Change management

As with all change programs, extensive repeated communication through a variety of channels is important to gain acceptance. The UK government met resistance due to lack of information where some believed relying on cloud providers would contravene policy and data protection rules.

7. Developing country issues

A recent report by UNCTAD2 identifies certain factors that are particularly significant to cloud decisions in developing economies:

• ICT is less prevalent in government operations;

• Some costs, especially access costs, may be more significant;

• Network topologies may not optimize the distance between the user and the internet;

• Cloud solutions may suffer more down-time and slower response in developing countries;

• Legal and regulatory frameworks may not address data protection concerns;

• Staff may fulfill certain processes more cheaply than automation;

• Applications that address local needs may not be available.

The UNCTAD Report notes that relatively few, larger emergent economies (e.g., Brazil, the Russian Federation, India, China, South Africa, Malaysia, and Thailand) have the scale and connectivity to support cloud data centers. Smaller nations could implement national data centers in their state telecommunications utilities and there is an opportunity to encourage the development of regional data centers located where there is already an established communications sector. Regional data centers may also mitigate some of the communications costs that would otherwise be incurred.

Conclusion

Cloud computing holds considerable promise for PFM and other public sector applications in developed and developing countries alike, and should not create a future digital divide. In addressing the challenges noted above, developing countries might want to seek technical assistance.

 

1 Chris Kanaracus, IDG News Service, www.cio.com May 01, 2014

2 United Nations Conference on Trade and Development, Information Economy Report 2013: The Cloud Economy and Developing Countries

 Note: The posts on the IMF PFM Blog should not be reported as representing the views of the IMF. The views expressed are those of the authors and do not necessarily represent those of the IMF or IMF policy.