Perspectives on Cloud Computing for Enterprises
* Faster transitions to new technology (due to shorter time-to-launch);
* Lower total usage costs when the right pricing tiers are selected (because the customer does not have to buy and deploy infrastructure to cover peak demand and expected growth);
* Improved resource elasticity and scalability, which allows customers to scale up or dial down services quickly in response to changes in business demand;
* Improved availability of enterprise applications like unified communications and collaboration services to mobile/remote workers;
* More efficient and effective management of technology resources by vendors with specialized skills; and
* Fewer technology management, maintenance and upgrade headaches (freeing internal IT teams to focus on more strategic initiatives).
Just What is Cloud Computing?
The definition of cloud computing is hazy, in part because many traditional IT vendors are positioning their products in the cloud. To qualify as cloud computing, a service should be:
* Hosted and managed by the vendor at its (or its supplier’s) facilities;
* Made available to customers remotely via an IP-based network (such as the Internet);
* Designed on a shared services/multi-tenant platform;
* Designed to provision resources dynamically and quickly in response to service requests submitted directly by users via automated, online tools;
* Set up to monitor and report usage in near real time; and
* Priced on a subscription basis where actual monitored usage is the dominant pricing variable.
Defining cloud computing is not just an academic exercise. Many of its benefits are available only when all of the characteristics listed above are present. There are three cloud computing service delivery models.
Software as a service ("SaaS")--end users access a standard vendor-developed and managed suite of software applications via a Web browser (e.g., Salesforce.com, WebEx, RightNow);
Platform as a service ("PaaS")--customer application specialists and programmers use a preconfigured computing platform to build and run custom applications within the cloud ecosystem (e.g., Force.com, Microsoft Azure); and
Infrastructure as a service ("IaaS")--customer network engineers obtain commoditized computing resources from the cloud in a manner that allows the customer to install, configure and run customer-provided software applications on the vendor’s infrastructure (e.g., Verizon CaaS, AT&T Synaptic Services, Amazon Web Services, RackSpace, GoGrid).
When enterprises decide to purchase cloud computing services, they face many business, legal and technical challenges. Some of these are common to other technology transactions, but several take on heightened significance in a cloud environment.
Investments in cloud computing have to be as future-proof as the IT world permits. To do that, enterprise buyers need to minimize dependence on proprietary vendor technology and buy services and systems that are open-standards-based, will operate with other systems they may adopt in the future, and will grow (in capacity and complexity) with the customer's needs. When possible, customers should avoid investing in an extensive set of application program interfaces ("APIs") for critical systems that work with only one cloud vendor's services--that strangles flexibility by tying the customer to a specific vendor.
From a customer’s perspective, solid service-level guarantees ("SLAs") are essential. Depending on the type of cloud service, key SLAs include those addressing:
* system availability/uptime;
* configuration change timeliness (including availability of newly ordered resources)--particularly for IaaS;
* application response time and transaction throughput--for SaaS;
* management portal/tools availability/uptime;
* incident response and problem-resolution times;
* service desk performance--particularly for end user adoption and satisfaction for SaaS;
* back up data success rate, and data restoration times where the vendor will host the customer’s production data; and
* customer satisfaction (and its importance increases as the deployment in the enterprise widens).
Meaningful SLAs should trigger material service credits when the vendor misses its performance thresholds. And the customer should not have to ask for credits--the vendor should proactively monitor its performance and automatically issue SLA credits when due. Solid SLAs can also be undermined by overly broad exclusions, including procedural hurdles and liability limits hidden in the fine print.
SLA best practices include escalating credits based on the severity, duration and frequency of service failures, and processes like the following to limit the recurrence of service level failures:
* requiring the vendor to conduct a root-cause analysis of service problems and to implement a corrective action plan at no additional cost to the customer;
* requiring the vendor to escalate chronic or critical service-level problems to its own and the customer’s senior management; and
* giving the customer the right to terminate the agreement for cause (or shorten the term) when performance failures reach defined levels.
For good reason, security is often cited as the primary concern for customers considering cloud computing. The vendor’s security management approach and tools should be evaluated closely. Security vulnerabilities in the cloud can be the result of design defects, poor patch/update management, ineffectual authentication controls, storing and/or transmitting sensitive data without encryption, and inadequate procedures for security incident monitoring, reporting and mitigation.
When evaluating security, consider:
(1) how the vendor addressed security in its design; (2) customer references;
(3) the vendor’s track record;
(4) the vendor's physical and logical security practices, processes and management approaches;
(5) the vendor’s commitment to continuously improve security; and
(6) the vendor's compliance (validated by independent third party audits) with recognized information security standards. Negotiate the right to conduct periodic security assessments and to assign responsibility for security incident detection, reporting, response and mitigation, and include a defined path for management escalation of unresolved security problems.
It is also important to be practical. Demanding levels of security that are not justified by actual risks or that far exceed the customer's own internal capabilities will get you little and lead to poor decisions. Data Governance
If the vendor will handle enterprise data, it should also demonstrate its data protection, disaster recovery and business continuity capabilities. The service agreement must clearly describe the vendor's obligations to protect the customer's data during its entire lifecycle, and the vendor should be liable for failure to do so. Contract clauses that support mature data governance include provisions that:
* Obligate the vendor to comply with the customer's vendor-facing privacy policies;
* State that the customer owns its data, has access to that data at its discretion, and will receive the data upon the expiration or termination of the agreement in a usable/standard format and medium;
* Identify the vendor's data extraction tools, and include the customer's use of them as part of the base scope of services;
* Describe the parties' responsibilities and liabilities for recovering lost or corrupted data;
* Define the countries where the vendor may (or may not) send or house the customer's data; and
* Require the cloud vendor to encrypt sensitive application data while in transit and at rest.
Technical, legal, and business issues face enterprises that want to use public cloud computing. Those that want to enjoy the cloud's benefits with minimum risk address provider lock in, service performance, security and privacy risks by selecting the appropriate service delivery model for each use case and negotiating service agreements (with capable providers) that get them the protections they need.
Joaquin Gamboa (firstname.lastname@example.org) and Marc Lindsey (email@example.com) are partners in the law firm of Levine, Blaszak, Block & Boothby, LLP. Joaquin will lead a session on Managed/Hosted UC at Enterprise Connect on Monday, February 28.