The metrics you need to gather to measure
SLA compliance depend on the level of service that needs to be supported.
These levels of IT support generally fall under the groups listed in Figure
1
- The level-one support group: This group is often the call center
or help desk. It should primarily monitor for application and
system status and service level compliance.
- The level-two support group: The responsibility of this group
is to handle basic problem resolution at the infrastructure level.
The level-two support group requires more detailed information
on the systems and applications than the level-one group does,
but avoid giving the level-two support group too much data.
- The level-three support group: This group is charged with detailed
problem resolution or capacity analysis, and needs as many metrics
as is necessary to capture pertinent data. Good level-three tools
employ artificial intelligence (AI) and sophisticated filtering
technologies to aid the troubleshooter in getting to the data
that really matters.
|
Listing 1 |
Ascending Levels of IT Support |
The Tools for Capturing Useful Metrics
According to DeAngelis, R/3 provides a control protocol called CCMS (Computing
Center Management System) that allows for tracking and alarming for a
variety of known errors and performance conditions. However, there is
much to track in a sizable R/3 environment, and the causes of errors are
not always easy to determine. You may have to use third-party tools, like
those from Envive, Luminate, realTech, and syskoplan.¹ Other companies
like Tivoli, Hewlett-Packard, or BMC also provide products that might
help in this area, but the products’ abilities should be checked thoroughly.
DeAngelis suggests five different sources
of service level data for R/3 that allow the capture of useful metrics:
- SNMP Mangement Consoles
- Intelligent Systems Agents
- Application Performance Monitoring Systems
- End-User Transaction Performance Monitoring Systems
- The ARM API
Details about each are provided in Figure 2.
While each of the tools mentioned here
can help with SLA maintenance, you should always first check the upcoming
features of the SAP system itself. Starting with Release 4.6C, the System_Administrator_Workplace
is available in the CCMS, which provides a view of your systems status
at a single glance. The new reporting functionality (RZ23) for the values
stored in the SAP Performance Database was also delivered with 4.6C. SAP
is working on these products to make them even more reliable in the next
releases, and the key transaction RZ20 (Monitoring Architecture) and RZ21
will also be redesigned.
SNMP Management Consoles
|
“SNMP” management consoles make good tools for
the level-one support groups because they are pertinent to health
and status monitoring. SNMP managers, such as HP OpenView Network
Node Manager or IBM NetView, use a simple but effective polling
architecturewith a management agent based on MIB-II to collect and
trend information on the availability of any device. Cabletron Systems
and Computer Associates offer similar SNMP managers in their management
tool suites. The data you extract from MIB-II predominantly concerns
real-time connectivity and proper operational status of the MIB-II
managed devices. The SNMP manager can also define and receive alerts
based on MIBs exceeding predefined variables.
|
Intelligent Systems Agents
|
Intelligent systems agents, useful to all three
levels of the IT support staff, run in the background on servers
and PCs. The agents used by Tivoli, HP OpenView VantagePoint, or
BMC Patrol are capable of tracking thousands of system health, performance,
and utilization variables. They derive this critical health data
from system log files, operating system processes, and other system-specific
resources. These metrics are used to set up alarm thresholds that,
when exceeded, generate alerts to the level-one support group, or
initiate preconfigured automatic administrative actions like rebooting
a dial-in server, which can greatly minimize delays.
|
Application Performance
Monitoring Systems |
With the preponderance of mission-critical applications
in recent years, many of the Monitoring Systems systems management
vendors have been investing in add-on extensions to their OS and
systems monitoring agents that add monitoring and threshold-based
alarm management to service metrics specific to R/3 application
modules. HP OpenView Manager for SAP R/3 features an API that links
to CCMS, and allows for R/3 programs to be monitored for trouble
signs. Events generated from R/3 events can be escalated through
pagers, cell phones, and e-mail, as well as tracked and graphed
from the Service Level Reporting module. HP OpenView Manager’s “Smart
Plug-in” is used in conjunction with HP’s VantagePoint Intelligent
Agent to supplement the existing OS and system monitoring capabilities
with additional knowledge on R/3. Envive’s Inspector monitor deploys
a multitiered management collection architecture on Windows NT or
2000. Inspector keeps a datamart updated with SAP R/3 vital signs
that are collected by the Inspector Server, which communicates with
the managed environment of R/3 servers. Envive has also been evolving
their R/3-focused management tools into a portal-based, “pay as
you go” strategy. Other tools in this space include Tivoli, with
its R/3 Manager, and BMC, with its solution based on Patrol.
|
The ARM API |
The ARM API Management via the “ARM API” is well
suited to the level-three staff because it emphasizes overall health
and performance, and filters out extraneous events and data, presenting
only relevant events with root-cause analysis of system faults.
In combination with application-related data derived from intelligent
agents, the ARM API can quantify the types of delays that really
affect the end user. Other tools are also coming to market from
Computer Associates, Candle, Hewlett-Packard, and other systems
management vendors that address the need for AI in determining the
root cause of R/3 performance degradation.
|
End-User Transaction Performance
Monitoring System |
End-User Transaction Performance Transaction (end-to-end)
monitoring tools monitor or simulate actual end-user transactions.
Monitoring Systems Monitoring tools like HP Web Transaction Observer
or Vital Sign’s VitalSuite place an agent on the client. For Web clients,
either ActiveX or JavaScript can be used. In either case, the end
users must accept the request for monitoring in order for their transactions
to be instrumented. Synthetic transaction tools like S3 from NextPoint
Networks (now owned by NetScout) and load-testing tools by Mercury
Interactive may be more appropriate for applications where users are
reluctant to be tracked. Companies like Vital Signs (now owned by
Lucent) have agents for non-Web applications like R/3. These agents
can be installed on the corporations’ computers, where users do not
have a choice about being monitored. Another good approach for R/3
is Envive’s StopWatch, which users a “blackbox” approach to R/3 end-user
transaction monitoring that tags and retrieves information from the
R/3 data packets transmitted from the clients to measure end-user
performance of the SAP GUI screens. With this information, StopWatch
can provide statistics such as bytes transferred per SAP GUI screen
session and average throughput for the different R/3 network segments.
Other tools in this area include Tivoli Application Performance Management
together with Tivoli TDS. |
Figure 2 |
The Five Sources of Service Level Data for R/3 |
You might also find necessary information,
such as end-user response time values, in the new transaction ST03N (since
4.6C), or in the Monitoring Architecture, where the values for transactions
are also available. You can also create alerts or, for example, send an
e-mail if a threshold is exceeded. This is all customizable in the R/3
system itself.
¹ Check www.sap.com/solutions/compsoft/cspdirectory for
SAP-validated products. To obtain more
information on vendors discussed here,
please see the “SLA Resources & References” section
at the end of the article. |
² Now owned by NetScout
Systems, Inc. |
³ Now owned by Lucent
NPS. |
The majority of this article is excerpted from an article
published in the November/December 2000 issue of the SAP
Professional Journal — “Defining SAP
Service Level Agreements: An IT Manager’s Survival
Guide,” by Richard C. DeAngelis, Jr. To receive
a complimentary trial copy of this issue, contact SAP
Professional Journal at sheila@SAPpro.com. |