This document was prepared in order to detail the mechanics behind a Service Level Agreement between a fictitious Company ABC and an outsourcing vendor. Several specific requirements have been assumed and documented in this article. You will want to replace the requirements of company ABC with your company’s specific requirements. My goal was to product a template that other companies could use to begin work on their own SLA.
There are two major parts to an SLA: the governing document and the process.
- The SLA Document is usually legally binding between a company and an outsourcing vendor(s). The document describes the exact services and service levels, with details about all agreements.
- The SLA Process represents the methods that the outsourcing vendor will use to support the SLA document. The methods of supporting the SLA document are usually left to the outsourcing vendor to identify. These processes should be discussed and possibly identified during SLA contract negotiation. It is important that both parties understand the processes and methods of support as well as the management and reporting tools.
The SLA process represents a third of the total solution. It is up to the outsourcing vendor and your company to ultimately choose the correct people to manage the systems and the best technology for implementation. The people involved in managing the process must also manage the technologies and understand the importance of reporting and monitoring the entire system.
System management and service desk automation technology can provide a supporting environment for tracking, escalation, and management of service metrics. End user satisfaction surveys can also provide input that will help target appropriate service levels and cost controls.
Service Level Agreements are often categorized in the following manner:
- Basic: A single level service agreement is in place. Metrics are established and measured, possibly requiring manual collection of data for management reporting. Objective is to justify the technical support operation.
- Medium: The automation of metrics data enables more comprehensive less labor intensive reporting of service level achievement. Introduction of cost recovery that maps to market rates and supported by service level reporting. Possible multi-level service agreements by cost per services rendered. Objective is to match service and cost levels with long term goal to increase service levels while decreasing costs.
- Advanced: Service levels are embedded in overall service desk processes enabling dynamic allocation of resources either external or internal to meet changing business conditions. Goal is to provide a seamless mix of services, costs and service providers at better than competitive rates. Often enterprises at this level are ready to extend services to the open market.
For purposes of discussion, the remainder of this paper examines the considerations of a company evaluator who must complete an SLA document of the support requirements for Exchange/Outlook systems. We will call this company “ABC Company.” The evaluator works with one or more outsourcing companies to negotiate the final agreements recorded in the document. Recommendations and suggestions, which are based upon industry standards and project management experience, are provided throughout.
The primary objective of the SLA document is to correctly identify the support requirements for Company ABC in regards to supporting the Outlook/Exchange infrastructure.
The ABC company evaluator alone cannot determine the appropriate details for the SLA. The outsourcing vendor’s industry experience and project management capabilities will provide required information and guidance. In many cases, it will be required for Company ABC and management within Company ABC to conduct workshops on the issues to determine specific objectives.
Moreover, we should all use our best judgment in collecting ideas and suggestions from the appropriate people. For example, for specific questions regarding helpdesk requirements, the outsourcing vendor may need to be involved in order to correctly identify a requirement unknown to the Evaluator.
Service Level Agreement Document
The processes in creating the SLA are broken down to ease the management of the project. The first four sections require the input of ABC Company management and, in some cases, end-user surveys. The next group of tasks may require the input of the current out-sourcing company in order to ensure all requirements have been identified. Next, the evaluator assembles the data in a document that can be easily read and understood.
The last sections take place during the negotiations with outsourcing vendors. Usually, a legal instrument will be created to bind both parties to a final Service Level Agreement. While the final SLA will be based upon the evaluator’s SLA document, it is likely that sections will be added or removed as negotiations dictate.
Contract Specifics and Context
Management of the SLA is a critical part of supporting end users. Before we can determine if objectives have been met, we must first identify metrics and the specifics of the contract.
Contacts and Role assignment
First, name the key contact to the Service Level Agreements and delegate SLA management tasks to others. Other contacts for the SLA include:
Exchange connection into other corporate systems
Management of 3rd Party Outlook/Exchange Development
Remote and Dial-in Access
The frequency and detail of reports must be identified as well. Reporting can then be further broken down into two techniques:
- Automated system reporting should be implemented in order to provide current and historical data. This data should be made available to the above named contacts on a regular basis. The methods for providing reports to the named contacts may include a secured website or electronic mail attachments. Hard copies of the reports may also be requested. The reports, for these contacts should be fully detailed reports with data analysis and a trend summary for the month. Moreover, historical data should probably be included.
It may be necessary for regional and divisional managers to receive a summary report/graphic depicting uptime and overall system performance once a month, similar to the graph depicted to the right.
ABC Company may also require that an automatic mechanism be put into place to notify the named contacts when critical performance thresholds are met. Specific thresholds are discussed later in the document.
Questionnaires and end-user canvassing methods should also be performed by ABC Company and/or the Outsourcing Company as part of an overall customer service initiative.
Payment terms and contract length are negotiated with the outsourcing vendor. ABC Company may prefer a contract length of six months, but will consider contracts as long as one year. Renewals can be handled in many ways including automatic six month extensions. Both ABC Company and the outsourcing vendor should be able to request a formal renewal meeting to update the SLA with riders and to negotiate new terms.
There are two types of terminations possible:
Contract Termination- Indicates that either ABC Company or the outsourcing company elects to terminate the contract. A “Technology Transfer” and associated fee would probably be required in order to shift the maintenance and support to another group.
Technology Termination- A termination in technology would occur when the support requirements are no longer required due to a shift in ABC Company technologies. This form of termination may or may not require a formal “Technology Transfer.”
Termination Options are described as follows:
- ABC Company may reserve the right to cancel the contract for either termination option with 60 days notice to the outsourcing company. ABC Company understands that there may be financial penalties for “Contract Termination” if the SLA objectives were met by the outsourcing vendor. These penalties often reflect the fee for one month of support.
- The outsourcing vendor may reserve the right to “Cancel Termination” with 180 days notice to ABC Company. A “Technology Transfer” fee would be charged to cover labor costs associated with transferring the knowledge and technology to another group.
There should be a formal review to evaluate the performance and customer service levels as well as staff reviews. A quarterly review is sometimes formalized in order to include discussions on SLA fulfillment, staffing and future projects that may affect the SLA.
Service Level Management is accomplished by negotiating a change or additional to an existing Service Level Agreement. Out-of-scope or new projects need not be discouraged. A change process occurs during every review process and can also be instigated as needed. Several things could require a change or addendum to the existing SLA:
A change in the process workflow
Missed performance or customer service thresholds
Additional third-party applications
Changes are not made directly to the SLA. Instead, contract riders are appended to the SLA until such time that the SLA is rewritten to incorporate the addendums. The SLA can only be written during a renewal cycle with both parties present.
Financial Incentive Plan
Most groups believe that the total cost of ownership (TCO) is more a function of cost of service and support of the system than a function of the cost of hardware and software. SLAs can drive down TCO by identifying damages for missed service levels.
In the case of ABC Company, a third party may be asked to provide evaluations to determine if service level objectives have been met. The costs associated with the third-party evaluations are the responsibility of the party requesting the evaluations.
Penalties and bonuses for SLA performance guidelines could be “paid” quarterly. Performance objectives are met based on a +10/-10 (percent?) allowance. Penalties are paid as a deduction of regular costs for the pay period immediacy following the review cycle. Bonuses are paid with four weeks of the review cycle and do not require a separate purchase order from ABC Company.
Performance Level Guidelines
Inter-site Message Transfers
Because the outsourcing vendor may have little control over the stability of the hub servers, ABC Company may not require guaranteed delivery times for mail originating from, or addressed to, any mailbox outside of the ABC Company’s site.
However, inbound Internet email with legitimate addresses should not get returned as undeliverable from the Exchange systems within the supported (environment?). The outsourcing vendor should remedy any internal Exchange process that returns mail.
Intra-site Message Transfers
ABC Company requires that intra-site Exchange mail be delivered to the recipient’s server-based mailbox within 15 minutes of delivery to a server within the supported site.
Remote Synchronization Performance
Off-line Address book
Remote users who replicate the Offline Address Book should never wait more than thirty minutes for a complete refresh to transfer over a 56K connection.
You should probably define the mailbox limits. In many cases, mailboxes are classified into two or more categories. For example:
50MB Users have a 50MB limit on mailbox size
75MB Users have a 75MB limit on mailbox size
Directory update frequency
Many companies configure directory replication so that the directory is current within a forty-eight hour time period. For example, a mailbox that is added at 3:00 p.m. on Tuesday must appear in the directory and Offline Address Book before 3:00 p.m. on Thursday. This should be defined in the SLA.
Administrative tasks, such as Exchange/NT username add/remove/change, should be able to be handled within one business day. The primary and backup responsibility may be divided up among the server support team and the helpdesk.
An matrix of administrative task groups and responsibilities should be created in order to identify the ability of each group, including:
- Add/Remove/Change of mailboxes and distribution lists
- One business day or less
- Public Folder Creation
- Two business days per request
- Exchange Gateway settings
- Two business days per request
- Updating permissions and security settings on Public Folders
- One business day or less
- Assigning email addresses to a Public Folder
- One business day or less
- Distribution List creation
- Two business days or less
- Distribution List modification
- One business day or less
- Mailbox restoration (from tape)
- Three business days or less
The outsourcing company may want to define the maximum number of one-day requests that can be filled per business day. Additional requests will roll to the next business day and will take priority over new requests.
Note: Public Folder applications that require or use automation, such as routing or scripting, are considered a separate project. New projects, which are likely to incur extra costs, are not part of the SLA.
System availability can be an expensive requirement. It is important that we identify the specific requirements from a resource access standpoint and not necessarily on a server by server basis. The specifics dictate the availability of the servers.
Network and remote access
Network connectivity between sites and for users should be defined clearly as to the required uptime. Redundant links may be required based on the connectivity requirements.
This specification details the amount of time a user cannot access his/her mailbox on an Exchange server in the supported site. Many companies define at least two types of mailbox classifications:
- Class A users can be without access to their mailboxes for no more than six business hours. This group usually contains managers and key people within ABC Company
- Class B users can be without access to their mailboxes for no more than 24 business hours. This group represents the bulk of the ABC Company Exchange users.
Public Folder Access
In order to maintain the OST consistency of applications, you should define how long Exchange users can be without a Public Folder. Some companies place this between 24 and 48 hours. However, you should request that, if restored, the Public Folders must contain all items as well as the most current items possible.
In the case of a corrupt directory, replication may need to be forced in order to get the names and addresses of ABC Company mailboxes outside the supported site. The site should not be without access to the complete directory for more than 48 hours.
In the case of a corrupt directory, intra-site replication may need to be forced in order to get the names and addresses of ABC Company mailboxes within the supported site. The supported site must not be without access to the complete directory list for more than 24 hours.
Information gathered from the previous specifications dictate the level of availability that is required. The services are then classified using the following availability classes.
While the table represents system availability, it is important to note that the figures represent unscheduled down time. It is critical that “windows” are allowed for scheduled maintenance and upgrades. The down time is always be scheduled on the same day every week over the weekend. Many companies detail acceptable times during the weekend such as 11 p.m. Saturday to 2 p.m. Sunday. The specific time needs to be negotiated.
The maximum allowable scheduled down time per week for the Exchange systems should also be defined. For example, you could specify an eight-hour maximum window during the week and 25 hours one weekend per month.
You may want to send out a user survey in order to determine the best time for scheduled maintenance.
The outsourcing vendor must balance the uptime requirements with the inevitable cost. The foundation for a mission-critical architecture has specifications for server availability, data accessibility, data protection and disaster tolerance.
Equipment Support Requirements
Access and Security
ABC Company should require that named contacts be permitted physical access to the equipment at any given time. Moreover, overall access to the equipment must be secured and restricted. Access to the equipment must be available 24 hours a day, 7 days per week for the vendor’s support personnel.
Disaster Recovery Preparation
Many companies require that the clients be able to request a “recovery of deleted items” for up to 30 days of deleted items. Moreover, backup tapes to the system should be placed in a 30 day rotation, then erased or destroyed. There should be no tapes that contain data over thirty days old.
In some cases, there may be a need to recover items from tape. The outsourcing vendor should honor such a request from any of the named contacts on the SLA. The outsourcing vendor may accept requests from the user community for restores, but should then verify the request with the named ABC Company contacts.
ABC Company does not have any specific requirements in regards to the types of systems (software) used to monitor the equipment.
Certification and Experience
ABC Company requires that at least one person supporting the systems have current MCSE status. At least two of the support personnel must be certified on the current version of Microsoft Exchange Server.
ABC Company requires that the resources that support the ABC Company Exchange systems be exclusive to ABC Company and not used for non-ABC Company projects or tasks.
While ABC Company has no requirements as to the brands or types of equipment used for the Exchange Server environment, ABC Company does require that the equipment be included on the Hardware Compatibility List for the current version of Microsoft BackOffice.
The outsourcing vendor is responsible for the requisitions and costs associated with all equipment necessary to support the ABC Company Exchange Systems.
Spares for testing/recovery
ABC Company requires that at least two entire servers be allocated for spare equipment and testing/recovery. The servers must match the current servers in production so that parts can be swapped and/or replaced. ABC Company further requires that the test equipment be updated as the production equipment and online at all times for testing.
Book- “How To Establish Service Level Agreements”, book, Naomi Karten
White Paper – “Negotiating An Effective Service Level Agreement – II” Gray, Gilbert & Tobin
Computerworld July 7, 1999 article “Building a better service-level agreement” by Deborah Radcliff
Book- ‘The Complete Guide To I.T. Service Level Agreements, “ –book Feb 6, 1999, Andrew Hiles
White paper – HP IT Service Management Reference Model
Network Management Forum, “Performance Reporting Definitions Document”, Issue 1, Apr 1997.
IETF Network Working Group, “RFC 1566 Mail Monitoring MIB”, N. Freed Ed, Jan 1994.
Book- “Service Level Agreements – Managing Cost & Quality in Service” Relationships, 1993, Chapman & Hall
Electronic Messaging Association Messaging Magazine January/February 1996, Clive Horton
Forbes Web Page – 9/7/1998 article “Killer Technologies Management Strategies” by James Champy
IBM Customer Solution Series Web Pages
Digital-Compaq Services, article, June 8, 1998, “Digital Introduces New and Enhanced Business Critical Services and Capabilities”
White paper – Proven Solutions Inc., “Measuring High Availability Power Protection Systems-The Power Availability Index”
White paper – DMR Consulting Group Inc., an Amdahl Company, “Managing Information Technology to Achieve Results”
White paper – “Negotiating an Effective Service Level agreement – I” by Anne Cain, Lawyer, Gilbert & Tobin
InformationWeek Dec 7 1998 article “Call Centers Evolve Into Customer Data Sources – Open Platforms Spur Service Expansion” by Mary E. Thyfault
Computerworld March 23 1998 article “Wal-Mart’s Help Desk Helps Itself” by Kim Girard
InfoWorld April 20 1998 article “Ten Ways to Improve your Help Desk (And They Don’t Cost Much Money)” copyright 1998 InfoWorld Media Group Inc.
Software Magazine July 1998 article ” The World’s Greatest Help Desk” by Rick Whiting
Computerworld July 6 1998 article “Help Desk Tools Gain Ground” by Kim Girard
PC Week August 3 1998 article ” Do It Yourself Help System” by Herb Bethoney
InternetWeek August 3 1998 article “Help Desks Can Boost IT’s Corporate Image” by Ralph Baskin
White Paper- “A Guide to Monitoring the Performance and Availability of Microsoft Exchange Server” 1999, Tom Kemp
InformationWeek Oct. 26 1998 article ” What to Look for in Metrics” by Noah Schactman
Computerworld Feb 23 1998 article ” An Investment in Uptime” by Laura DiDio