the 10 reportsmedia.techtarget.com/searchenterpriselinux/... · • improve performance by managing...

24
01 . 02. 03 . 04 . 05 The 10 Reports... Every Information Technology Manager Lives For THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR Free Guide for Managing the Enterprise: UNIX, Linux, and Windows

Upload: others

Post on 20-Jul-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01.02.03.04.05

The 10 Reports...➤ Every Information Technology Manager Lives For

➤ ➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤

Free Guide for Managing the Enterprise: UNIX, Linux, and Windows

Page 2: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

Business leaders today are driving IT initiatives to capitalize on Internet technologies and

web-based application delivery. They understand that web-enabled applications are a fast,

effective way to expand the reach of existing systems, enhance customer service and

deliver internal solutions. Web-based application delivery is seen as a driver to a competitive

advantage, as well as an effective way to reduce costs and enable new revenue.

As such, your infrastructure has grown into a complex collection of UNIX, Linux and

Windows servers using web servers to connect end users with browser technology and

web application servers to connect to back-end databases and legacy applications. This

complexity is made even more demanding with the introduction of today’s .NET and Java

technologies, which promise the ultimate in rapid-delivery, cost-saving application deployment.

While web technologies deliver the performance, flexibility and integration you need, they

also add new levels of complexity. These new architectures derive their flexibility by adding

a variable number of loosely coupled distributed elements ranging from transaction

management and database connection pools to components (servlets) that exist in an

ever-changing mix, growing and shrinking as system loads increase and decrease.

These web application technologies promise to deliver exactly what users want—

ubiquitous application access, anywhere, anytime.

Now, all you have to do is manage it.

Managing Your Web ApplicationsAnytime, Anywhere

Page 3: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01.02.03.04.05.06.07.08.09.10.01.02.03.04.05.06.07.08

What NetIQ Can Do For You

NetIQ’s Performance and Availability Management solutions can provide insights into

how well your web-based application infrastructure is serving your end-user population.

NetIQ analyzes and reports on end-user response time, web servers, web application

servers and back-end databases.

The Reports You Need

You have questions about managing your web-based application infrastructure. This

guide has been created to show you 10 critical system views and management reports

that can help answer your most pressing questions. Discover how this information can

be used to optimize the performance and availability of your infrastructure to improve

user experience, exceed Service Level Agreements (SLAs), save time and increase the

return on your infrastructure investments.

Page 4: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

01. What is the end-userperformance experience?

The best measure of success is the end user’s experience

All IT efforts are wasted if a mission-critical application is unavailable or performs poorly. Monitoring

the end-user response time of web-based applications is essential as more and more of the enterprise

accesses applications, processes transactions, performs vital business functions and keeps in touch

with the virtual organization through the Web.

Looking at web-based transaction performance and availability as experienced by a typical end user

enables you to deliver service levels based on the business processes that are vital to your organization’s

success. By recording and playing back key business transactions from any location, you get consistent

and repeatable metrics on availability and performance. This data supports optimization, speeds

problem characterization and lowers support costs of your web-based application infrastructure.

What can you do with this information?

• Manage web sites at the business process level by enabling your IT staff to see the performance and

availability that users are experiencing.

• Lower support costs by identifying the response time of all supporting infrastructure for key business

transactions from the end-user browser through the network, web servers, middleware and

application servers to the back-end database. You can easily identify the infrastructure components

responsible for any degradation.

• Optimize site performance to meet and exceed SLAs for maximum Return on Investment of your

web-based application infrastructure.

01 02 03 04 05 06 07 08 09 10

Page 2

Page 5: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 3

Response Time of the Customer Account Balance Transaction

This chart shows the response time of a key business transaction over time. It highlights graphically when response timesexceeded the business service level and which steps were responsible for those exceptions each time.

Page 6: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Web servers are the front line in the new world order of IT

With more and more users accessing organization data, personal data, news and applications via the

web front-end, a robust web server is essential to every enterprise.

Basic load is always a figure of merit for a workhorse application like Apache. Many factors drive

load on a web server, from graphic-intensive pages and extensive processing to external factors,

such as a news event that creates unexpected demand. By looking at the computational load

generated by all the virtual servers running under an instance of Apache, we are able to watch this

metric closely.

What can you do with this information?

• Prevent outages by alerting on above-normal loads that could overwhelm the web server.

• Identify the sources of high CPU demand so you can predict future loads, plan upgrades and

schedule maintenance during periods of lowest load.

• Track trends of CPU utilization over time to understand long term trends in load versus normal

spikes in load level.

01 02 03 04 05 06 07 08 09 10

Page 4

02. Is the web server healthy?

Page 7: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

02CPU Load of the Web Server

➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 5

The data shows a relatively variable load, moving to a high level of CPU utilization and back down again. During peak timesthis is a very busy web server, so our threshold of 38 percent CPU is correspondingly high. With this load level, it would beimportant to examine the same data over days and weeks to ensure that no underlying trend is pushing this web site intoeven higher load.

Page 8: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

On the Web, speed is key

Responsiveness seems to be the primary quality by which users judge everything related to the Web.

In addition to the user experience view from outside the data center, the web server can provide

metrics about its ability to respond to requests.

Monitoring both the average and maximum processing time of recent requests provides a balanced

picture of performance as it impacts the user. With this information, you can make sure your web

server is responding with appropriate speed.

What can you do with this information?

• Ensure high availability of your mission-critical servers by identifying responsiveness issues before

degraded performance drives away customers.

• Optimize web server performance to meet IT service level commitments to the business.

• Verify that pages are being served in a timely way, so your users and customers are satisfied.

01 02 03 04 05 06 07 08 09 10

Page 6

03. Is the web server respondingfast enough?

Page 9: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

03Web Server Response Time Average Versus Maximum

➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 7

In this example, the average processing time ramps up at 9 p.m., when the server is brought into the production loadbalancing rotation. Although occasional longer processing times can be seen prior to that, once the production load hits theweb server, the maximum remains quite stable while the actual average fluctuates. Around 9:30 a.m., a brief network outagecauses requests and processing time to drop, resetting the maximum value. As soon as that event ends, both return to theirnormal production levels.

Page 10: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Know that your enterprise data is protected

Every enterprise has backup and recovery utilities that offer secure protection of distributed

mission-critical databases. Knowing the status of these for each database not only lets you optimize

performance and ensure availability through automated event detection and correction, but also

gives you the knowledge that your data protection utilities are providing the security they promise.

A single view of completed, failed and pending data backups provides an essential overview of your

backup processes. Viewing this information alongside system resource information, such as memory

and CPU utilization, lets you know how your backup processes are affecting performance.

What can you do with this information?

• Manage consolidated information more easily and quickly through a comprehensive view of the

status and performance of your automated data backup and recovery utilities from a single console.

• Lower support costs by viewing the health and status of your distributed data protection environment

from one central location and providing prepackaged business rules for managing the availability

and performance of the environment.

• Perform capacity planning and trend analysis with in-depth views of performance statistics—

statistics not available through standard management tools.

01 02 03 04 05 06 07 08 09 10

Page 8

04. Are my mission-critical data backuputilities working correctly?

Page 11: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

04Data Backup Completion

with CPU and Memory Utilization➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 9

Not only can you see completion status for data backups, but also critical performance information. One consistent viewdelivers the information you need to ensure that your data backup utilities are working correctly. In this example, backup jobsare being consistently completed without creating unusual spikes in memory or CPU utilization.

Page 12: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Make sure new users aren't exhausting system resources

With any user-access-intensive application, a target for optimum performance is to return

user-requested information from memory rather than by directly accessing the physical hard disk.

Performance problems are usually the result of contention for, or exhaustion of, some system

resource. As new users come online when a system resource is exhausted, the system is unable to

scale to higher performance levels—let alone maintain the current performance level. Careful system

management is required to prevent system resources from becoming exhausted, causing poor

service or downtime.

An application can be deemed un-scalable if it exhausts a system resource to the point where no

more throughput is possible when workload is increased. Such applications suffer in static

throughput and poor response times as user load increases. To understand the impact additional

users have on your database, you need to look at the number of users compared to critical

performance metrics.

What can you do with this information?

• Monitor SQL Server, Oracle and DB2 database RDBMS servers in a real-time or historical perspective.

• View the health and status of your distributed UNIX and Windows relational database servers from

a central location through pre-packaged business rules, known as Knowledge Scripts.

• Improve performance by managing how an increase in the number of database connections directly

impacts database performance, revealing whether database configuration or tuning changes are

required as more users connect to the database.

• Determine the effectiveness of database memory stores.

• Manage and automatically respond to database size and growth requirements and service outages.

01 02 03 04 05 06 07 08 09 10

Page 10

05. What is the impact of additionalusers on my databases?

Page 13: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

05Increased User Connection Impact

on System and Database Resources➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 11

Performance can be directly impacted by the inability to return user-requested data from memory, requiring disk I/O. Thisview shows a direct correlation between users and memory/disk utilization. In this example, the exhausted system resourceidentified is the database buffer cache. As more users come online, information in the buffer cache is used less and diskread/write access time skyrockets.

Page 14: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Give your databases room to grow

One of the keys to ensuring high availability of mission-critical database applications is to make sure

there is enough allocated storage available for the database to grow as needed. Many database

structures grow dynamically, making adaptive baseline usage and predictive analysis of storage

allocation difficult. Additionally, management is required to ensure database configurations are

accurate for storage allocation growth and access requirements moving forward.

Having storage allocation available for databases means much more than just having plenty of disk space.

There are also database configuration management disciplines required for the high number of database

objects to ensure the database will scale, extend and perform as required, regardless of pure disk

space. All of the individual components and data structures must be maintained on a continual basis.

Historically, up to 30 percent of the database administrator’s time is spent on storage allocation

management. Automated management of your database storage allocation lets you proactively

detect problems that can impact database availability.

What can you do with this information?

• Manage data, block, extent, segment, tablespace, file, log and database storage allocation for

Oracle, SQL Server and DB2.

• Proactively detect potential storage allocation problems that can impact your database system

availability and use automated corrective actions before users are impacted.

• Eliminate embarrassing web site and application errors that refer to database storage allocation

problems, such as ‘Unable to allocate next extent.’

• Determine the effectiveness of database memory stores and dynamic memory store configuration

parameters.

• Maximize database performance by ensuring that it is not compromised by default database

storage allocation parameter increases.

01 02 03 04 05 06 07 08 09 10

Page 12

06. How do I ensure there is enoughstorage allocated to my databases?

Page 15: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

06Database Storage Allocation Used

and Available Tablespace ➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 13

This view shows critical database storage availability. Once the threshold for tablespace usage was crossed, a corrective actionwas automatically taken to extend the datafile and add space to the tablespace without human intervention—avoiding a 7 a.m.phone call to IT because the database went down.

Page 16: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

J2EE Application Servers have a lot of moving parts

The first question you have to answer is the overall health and performance of the entire

application server.

A few simple metrics, like disk activity, requests and CPU load, paint an immediate picture of the

application server performance level. Many of the UNIX management modules have a predefined

“Health Check” Knowledge Script that watches basic availability metrics, tests the ability of the

managed application, such as the IBM WebSphere Application Server, to respond to a connection

request and alerts you to any fundamental problems.

What can you do with this information?

• Ensure that your mission-critical applications are available and performing optimally.

• Maximize uptime by responding to automated alerts of resource issues.

• Manage your complex application servers more easily with an overview of their general health and

performance.

• Detect potential problems around your application servers before they result in downtime.

01 02 03 04 05 06 07 08 09 10

Page 14

07. Is the application server healthy?

Page 17: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

07General Health of the Application Server

➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 15

In this view of our WebSphere Application Server, we see a spike in the maximum processing time for requests cominginto the server, which is accompanied by spikes in disk reads and disk writes and a modest spike in CPU. Overall, theCPU and other data show no significant trend, so the spikes are probably not indicators of a developing problem. Thedata shown might suggest that a lower threshold is possible, providing earlier warning of potential problems related toactivity bursts.

Page 18: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Tune your J2EE application server for optimal performance

J2EE Application Servers have a wealth of parameters that typically must be tuned in production.

Only operational monitoring can tell you how the entire server is working and highlight the bottlenecks.

Two possible suspects for performance problems are the JDBC (Java Database Connectivity) connections

and the transactions shown in the JTA (Java Transaction API) data.

JDBC is the way that the server gets data to and from the database. The JTA is a standard

mechanism for J2EE servers to do their work. To determine if your application servers are tuned for

optimal performance, look at key values including the total transactions committed, the total

transactions rolled back, the total transactions (the sum of the previous two items) and the average

commit time for transactions.

What can you do with this information?

• Ensure that your application server’s parameters are tuned for optimal performance through

operational monitoring.

• Identify performance bottlenecks between different kinds of resource allocation.

• Contribute to rapid client response times and optimal user experience.

• Detect and correct performance problems before your business is affected.

01 02 03 04 05 06 07 08 09 10

Page 16

08. Is the application serverperforming optimally?

Page 19: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

08JDBC Client Resources • JTA Transaction Profile

➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 17

The JDBC data contrasts the peak number ofclients waiting with the actual number ofclients waiting. In this example, taken shortlyafter the server was started, the peak rises tofive clients waiting and remains stable there.After a few minutes, during which the serverand back-end databases reach their steadystate, the number of clients waiting—whichnever exceeds two or three—drops offcompletely. If necessary, a smaller JDBCConnection pool could be allocated to saveon resources within the server.

The JTA data shows a lot of variance in the number of transactions handled,which is probably reflective of the demand. The number of transactions rolledback remains small and, even more importantly, does not trend upward.Basically, this is a picture of application server health. Only the small upwardspikes in the average commit time warrant possible additional investigation. Ifrollbacks were to become more prevalent, additional Knowledge Scripts couldbe run to provide a breakdown of why transactions are not committing.

Page 20: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Establish baselines based on a comprehensive set of metrics

Just as cars today use computer chips to monitor engine components, IT operations must be

prepared to monitor standard metrics and, when needed, collect deep diagnostics. NetIQ agents are

designed to perform routine threshold checks and integrate with operations consoles for event

notification and action.

Using a lightweight, unobtrusive design, a single base agent resides on each server. The agent uses

managed objects (MOs)—you can think of MOs as plug-ins to the agent—to monitor specific application

servers, middleware, data servers and other components. By centralizing scheduling and communications

in the agent, policies administered by MOs can be set and changed immediately, without changes to

the agent itself or impact on servers. Change is simple—just select from a list and deploy from a console.

NetIQ agent technology enables IT management to provide consistent service levels and exceptional

reporting. The agents provide a common set of metrics to establish baselines and set thresholds and

are easily modified to capture deep, detailed diagnostic information. Portable, powerful, flexible and

adaptable—NetIQ agents are technology for the e-business world.

What can you do with this information?

• Get a detailed view of the performance and availability of both Windows and UNIX servers with

statistics that go well beyond those provided by PerfMon counters or UNIX system utilities and

commands.

• Detect potential problems that could impact the availability of your most important systems, so

critical business processes continue to deliver the service your customers expect.

• Lower support costs by viewing the health and status of distributed systems from a central location,

tracking trends in performance and availability and automatically correcting problems.

• Easily set up event-driven actions—such as taking corrective actions, sending an e-mail or pager

alert or generating a SNMP trap to an enterprise console.

01 02 03 04 05 06 07 08 09 10

Page 18

09. What is the baseline performancefor a healthy server?

Page 21: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

09Baseline Performance of Server Resources

➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 19

Using agents to perform threshold checks, you can get a detailed view of which processes are consuming memory andincreasing CPU utilization. A collective view of healthy processes can be used as a performance baseline. This is an exampleof a healthy server, with CPU utilization only occasionally and briefly crossing the threshold, and a very small number ofzombie processes.

Page 22: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

01. How can I manage more serverstoday than I did yesterday?

Web server components are on the front line of application delivery

Application delivery is dependent on critical components of your web servers. When performance

degrades, you need to know the cause. Looking at resource utilization by application and server can

show which components or processes are causing problems.

High system resource utilization can be a source of performance degradation. Monitoring possible

runaway processes and the top processes consuming CPU lets you know if a critical component of

your architecture is causing a problem.

What can you do with this information?

• Save your staff time by automatically pinpointing abnormal resource utilization. An individual

server can be identified as the root cause of a performance bottleneck.

• Empower your junior staff to accomplish tasks normally reserved for senior personnel, reducing the

firefighting load to free up your more experienced resources to complete scheduled projects.

• Improve business performance by delivering the metrics that enable optimization of web server

resource utilization.

01 02 03 04 05 06 07 08 09 10

Page 20

10. Which components of my webapplication architecture arecausing problems?

Page 23: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

10Resource Utilization by Application and Server

Possible Runaway Processes➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤ ➤

Page 21

This chart shows one process—a runaway Apache web server process—exceeding a normal baseline and consuming most ofthe Red Hat Linux server processor. Normally, even at the busiest time of the day the baseline utilization should be 80 percent.Identifying runaway processes is critical to maintaining performance.

Page 24: The 10 Reportsmedia.techtarget.com/searchEnterpriseLinux/... · • Improve performance by managing how an increase in the number of database connections directly impacts database

© 2002 NetIQ Corporation, all rights reserved. RP10475PAM MH 1002 (p)

ContactsWorldwide Headquarters

NetIQ Corporation3553 North First StreetSan Jose, CA 95134713.548.1700713.548.1771 fax888.323.6768 [email protected]

NetIQ EMEA+44 (0) 1784 [email protected]

NetIQ Japan+81 3 5909 [email protected]/japan

For our offices in Latin America & Asia Pacific,please visit our web site at www.netiq.com/contacts

06.07.08.09.10➤ ➤ THE 10 REPORTS...EVERY INFORMATION TECHNOLOGY MANAGER LIVES FOR ➤

For more information about NetIQ’s Performance and AvailabilityManagement solutions and for free product trials, visithttp://www.netiq.com/solutions/performance/default.asp