What’s a service?
There are a number of ways to define service. But the easiest and most rough-and-ready is that a service is what the end-user experiences, and the user’s experience of that service determines if the service is working or not. An email service is the ability to send and receive email, the ability to log into the email system and see new messages and existing messages, the ability to look up a user’s information in the address book, and so on. For an order entry system the service is the ability to create an order, to look up customer information, or to price a quote.
This definition is fine for user-oriented services, but some services are only used by other services. For example, Active Directory is typically not accessed directly by users, but rather by other applications (such as messaging) that users interact with. For example, Exchange can be considered a user of Active Directory. However, even in this case, the service is defined not by the IT concept of the Active Directory servers, but rather by whether Active Directory responds quickly and correctly to requests for directory information.
For the purposes of automated monitoring, most relevant services are, of course, implemented using a collection of IT elements working together.
Service level management
The user community of a service typically has a set of expectations the service’s performance, availability and quality. Often these expectations are formalized in a service level agreement (SLA) between the user community and the IT organization. To be successful, IT must manage the service in such a way to meet these service level agreements or expectations.
Service level management involves measuring the user’s experience of the service to get real-time and historical data. It also involves using that data to ensure the service and its underlying IT components are working, diagnose it if there is a problem and provide reports to show compliance with service level agreements or service level expectations. Service level management activities include:
• Measuring service quality from the end-user perspective
• Using metrics that represent customer experience
Typical metrics for service level management are availability of the service (can I log into the SAP system?) and response time (can I load the HR benefits site web page in less than 6 seconds?).
• Using SLAs to articulate target service quality required by customers
• Evaluating the service on a per-service, per-customer basis
Systems management
Systems management is the familiar activity of monitoring servers and other IT infrastructure to detect existing and potential problems and notifying IT personnel when a problem is detected. Systems management is focused toward the IT elements that provide services, such as servers, applications, computer hardware, switches and networks. The activities in systems management include:
• Monitoring the health of individual elements and applications
Typical systems management metrics are CPU, memory and network usage; application errors; and connectivity.
• Rolling element health up to service operational health
• Identifying failures which may lead to or cause service quality impacts
• Evaluating on per-server, per-application basis
Service mapping
Assuming that we have both service level management and systems management in hand, there’s still one piece missing from the complete puzzle. It goes back to the need for IT to be able to take responsibility for both the IT elements underlying the services as well as the services themselves. Associating the customer experience with what’s going on within your IT systems (inside your server room, so to speak) enables you to do a superior job of managing the IT resources.
It’s clear that doing monitoring from a user perspective provides good insight for the IT manager. But what is to be done if a problem is discovered from the user perspective? At this point the IT operator knows there may be a problem somewhere in the data center, but where is it? This is where the service map comes in. One of the key values provided by a service map is to show how service level measurements—in particular, service level problems—relate to IT elements.
For example, a service map may show that end points in Houston, Tokyo and Munich are all focused on monitoring user activities against the main Exchange site in San Jose. If Houston, Tokyo and Munich all start reporting availability problems, then the operator, referring to the service map, knows to look in San Jose for issues. Chances are the Exchange site in San Jose has already reported some problems via the element monitoring occurring on its elements—the service map provides a simple way to correlate problems detected inside the data center with user impact outside the center.
Service maps have additional benefits. For example, if two Exchange servers in San Jose are showing problems according to element monitoring but no users (as represented by end points in Houston, Tokyo or Munich) are affected by those problems yet. The service maps show that one Exchange server provides email for the organization’s executives, while the other email is a gateway for the Japanese branch office—and it’s 3am in Japan. Using this information it’s easy to prioritize the fixes for the Exchange problems—fix the executives’ server first, then the branch office gateway.
If you have service level agreements with your customers, and a formal service catalog, you would have specific obligations for service levels to these two organizations, and could use that information to prioritize your activities.An example of service management close to home
Your home telephone is a good model for understanding a service and the interaction of service level management and systems management. From the perspective of the telephone company, the telephone system is provided by an extremely large, complex and resilient distributed system of computers, network devices, cables and software, as well as personnel like operators and supervisors. But from the perspective of the customer, there are a few basic expectations that define whether the telephone service is working: 1) when I pick up the phone, I hear a dial tone, 2) I can dial a number and get (and stay) connected, 3) if someone calls me, my phone rings and I can talk to them.
As far as managing the telephone service goes, the telephone company spends a lot of time and effort making sure the complex distributed system of computers, network devices and cabling can meet the expectations of its customers. This involves monitoring each of the components so that problems can be fixed before they affect customers, as well as using redundancy in the systems so that problems can be routed around while they are being fixed, and other techniques. But you can also be sure that the telephone company is “picking up the phone” on a regular basis, from many different locations, to ensure that what the customer gets is what the customer expects.
Furthermore, most states have public utility commissions, representing the users of the telephone service, that enforce strict service level agreements with telephone carriers that are analogous to the service level agreements IT enters into with its users.
No comments:
Post a Comment