Practical reports and applications
For nearly every organization whose business processes rely on IT, there is a need for network and application performance management.
The special challenges of IT management, which have a direct impact on performance management, include:
- A large number of different applications
- High sensitivity to availability and performance
- Large impact of many applications on operational productivity
- Increasingly rapid changes (mergers, centralization, IT integration)
- limited IT resources
- Open systems in the direction of external users / partners
Why is performance management an indispensable part of the IT management toolset?
- Avoiding operational and productivity losses
- Support for complexity
- The need for collaborative problem-solving processes (internally and with external partners)
- Appropriate handling of changes
This whitepaper is based on many years of customer feedback and represents a selection of application scenarios in which Performance Vision is used by SecurActive.
Challenge IT performance
Enormous application diversity
1141/5000 The number of applications running on the company's own infrastructure is increasing: each has now become double-digit applications, and larger organizations can go into the hundreds.
Each department now has special applications that support its operation. In some cases, individual departments process the same operations with applications from different manufacturers.
The result is an infrastructure that is used by a very large number of applications. Accordingly, it is difficult for IT administrators in the infrastructure sector to get a complete overview of how each application runs on the network. Furthermore, many manufacturers are highly specialized and relatively small companies. Their technological basis is sometimes no longer up-to-date, in some cases even inadequate with regard to new implementation scenarios (WAN provision, WiFi, remote access, etc.). This diversity culminates in an extremely high complexity for the infrastructure administrator.
High sensitivity to availability and performance
By definition, many organizations run 24 hours and 7 days per week without exception, and all business operations must be available at all times.
Time is a critical factor in providing products and in terms of the ability to support processes that in turn support your services.
Many applications have a great impact on your employees:
- Einige Applikationen sind geschäftskritisch und Ausfälle oder Fehlfunktionen lassen sich unmittelbar in Form von Umsatzeinbußen messen.
- Andere, weniger kritische Anwendungen haben direkte Auswirkungen auf die Produktivität Ihrer Mitarbeiter und ragen weit in kritische Abläufe und Prozesse hinein.
Increasingly faster changes
Many companies have quickly introduced computer applications to support many processes. Very few processes are still independent of IT applications today. IT and business operations are now closely interrelated, which has led to great productivity gains but also to high dependency on the IT infrastructure.
The need to open up and share common systems
Companies must now have the ability to make information available outside their corporate network. For example, resellers, dealers and customers who are based in other locations need access to deal with the company.
Limited IT resources
Despite the increasing support of business operations through computing and general growth, as well as the merging of companies, the size of IT administration teams has hardly changed.
What will bring you performance management?
Performance management solutions support IT:
- to recognize the performance as perceived by the end user
- find the cause of speed losses quickly
- understand how their infrastructure is used
- as well as to continuously improve overall performance
Given the complexity of the IT systems and the available resources, N / APM solutions are needed that support the IT infrastructure team in providing the right foundation for business operations.
Organizations that implement network-based N / APM solutions typically undergo the following four phases to significantly improve performance with their N / N / APM solution:
Immediately after implementing the N / APM tool, the teams usually begin collecting current performance data from their IT systems. They identify a list of problem areas that have accumulated over time and various IT migrations over the past few years. Some of them affect the end users, some do not, or not.
Within a few days they determine:
• The applications with the worst performance
+ A transparent presentation of the applications with the worst performance from the end user's point of view. + Clear view of the origin of deceleration / insufficient performance (server, application, network, etc.)
• Performance gaps within the network
+ Which segments are affected?
+ For which applications?
• Name resolution problems
+ DNS / WINS server performance
+ Incorrect DNS targets
+ DNS configuration on system level (eg no cache)
• Incorrectly configured hosts / servers
+ Systems with an inherited inadequate configuration
+ Poor network configuration
+ Blocked network traffic
• Abuse / bandwidth
+ Unnecessary network usage (unauthorized applications)
+ Consumption-intensive applications
+ Processes that are in unfavorable timeframes (eg backups).
After this initial assessment, the IT infrastructure team has typically provided a clear picture of a whole series of problem areas, and is thus able to create a design to improve overall IT performance.
Accelerated problem solving and reduction of operating losses
Once the N / APM solution is ready for use, it should reduce the time that end users are affected by poor performance. The cost of performance degradation is directly related to the speed at which the problems are resolved (for more information, see the following whitepaper: "The cost of performance degradation").
The solution should be able to influence MTTR (Mean Time To Resolve) in various stages of the problem solving process:
Improve response times
Organizations that do not have a performance management tool for their network and business critical applications are informed by end users about the issues.
This information source is very slow because:
- Users usually only call when they are angry or desperate.
- Because the performance statements are based on the user's subjective feedback, they are not a reliable source of information from the administrators' point of view, as long as no large number of users are affected or the problem is escalated.
- User feedback is too inaccurate or not sufficiently specific to allow the administrator to diagnose the problem. This is often confronted with statements such as "the network is too slow," where clear information about the problem, such as the affected application, the time of the disruption, which transaction was currently running, and exactly when it was executed, etc. would be important to him.
- Performance bottlenecks often occur only on a temporary basis, that is, even when they are reported, the administrator can not analyze what has just happened without an adequate tool.
An N / APM solution reduces reaction times in many ways:
- Administrators are informed of the most serious limitations before users reach the phone.
- Administrators can rely on objective, unambiguous metrics to evaluate the Quality of Experience and clearly identify deterioration.
- Administrators have a detailed information history of the incident and can start the diagnosis.
Find the origin of the problem
At this point, most enterprises that do not have any tools fail: performance perception is subjective, the problem only occurs in a timely manner, and traditional monitoring tools focus exclusively on availability and resource consumption (and do not track performance) , This means that most IT departments in their environment (network, server, database, etc.) can not find anything extraordinary and then limit themselves to the proof that the infrastructure section for which they are responsible is not responsible for the problem.
Until the problem is escalated and a clear decision to analyze the problem may take several days or even weeks. During this time, valuable productivity is lost on the user's side, enterprise productivity drops, and frustration is spreading on both the user and the administrator side.
A performance management solution follows the development of the Quality of Experience and also provides a user-friendly interface to display using a set of metrics that allow the administrator to answer the following questions with a single click into the main dashboard:
- Is there a deterioration in performance?
- For which applications?
- For which users?
- Where is the cause of the deterioration? (Network, server, data volume, etc.)
A few additional clicks allow the administrator to view the activities and transactions of a single user in detail. Once you have automatically passed all this information to all (internal and external) stakeholders, the joint problem-solving process can begin, with each department taking over the part that falls within its competence.
Determine the cause
Once you have access to the information above, and your performance editor automatically makes a record of the slow transactions, it becomes easy to determine the cause. To gather the same information, you would have to evaluate protocols, configuration files and graphical analyzes from traditional monitoring solutions in order to get to the root cause.
We received a screenshot of a customer who wanted to share his experience with us. A lot can be learned from this. We can see exactly how the Quality of Experience has evolved for a given application during the day.
- It is clear that the server response time (blue) has deteriorated more and more over the course of the morning and only a limited number of transactions have been processed.
- at 12:15 pm the application crashes.
- After 15 minutes the application restarts and the response time of the server 8In Blue) normalizes. There are now significantly more transactions processed, probably to catch up those that could not be processed in the morning. Our customer has reported to us that the application ran a "suboptimal" RAM usage by keeping any file opened by a user and then not closed (and it seems to be a large number of files to go) in the RAM stored. As soon as a certain number was exceeded, this was no longer possible, the server began to transfer the files to the hard disk, thus slowing down.
Once the organization has the ability to deal with performance degradation, the next step is to protect users from restrictions. A performance management solution helps by focusing the attention of the administrator on potential problem areas and supporting it in improving performance even if there are no user complaints. In fact, users take the initiative only when the poor performance seriously affects their productivity. It is important to seek improvements for particularly critical business applications to contribute to the efficiency of the company.
Performance management solutions provide you with the following:
- Track errors / problem areas within your infrastructure
- The actual performance levels and drivers
- The ability to communicate clear and unambiguous data on performance
A performance management solution reports problems, even if they do not result in user side complaints: configuration errors, such as systems that call a primary DNS server that does not respond, lacking a system's cache, non-availability of a system in An application chain, bandwidth guzzlers, network capacity abuse, etc. It is likely that these problems are currently causing a slowdown. In combination with other problems, they can be the cause of future performance degradation. By identifying these problems, a performance management solution can help prevent future losses.
A performance management solution also reports trends and performance drivers of individual applications. Based on this, intelligent adaptations are possible, which proactively improve the performance of this application.
Excerpt from the presentation of a software developer who describes the application improvements using Performance Vision.
On this slide, the software manufacturer explains that the elapsed time in the server response time (blue part of the cumulative graph) has dropped nearly 800 ms. The server performance is now much more stable (see second graph) and rarely exceeds 60 ms (with increased application activity - from 12,000 to 16,000 transactions at peak times).
(Note: This graphic shows three colors: blue for server response time, pink for data transfer times, and yellow for response time of the network.
Excerpt from the presentation of the software producer, which represents the reduction of the average server response time (SRT).
On the second slide the software house shows the reduction of the average server response time from 160 to 40 ms. According to our customer, the feedback of end users has already been proven to improve.
Companies see themselves exposed to a high number of rapid changes from year to year. These may differ in size, but always involve a certain risk with respect to the performance level for the end user. Changes that affect the IT team range from the simple provision of new application versions through application migrations to the merger of IT landscapes.
To minimize the risks associated with such changes, IT administrators need to evaluate the current state of their infrastructure (efficiency, current problems, network and application performance), the impact of the changes, and control the real impact during and after migration. Lack of transparency leads either to performance degradation or to an over-dimensioning of capacity and thus to a waste of financial resources.
How can you select your performance management solution?
Various performance management approaches
There are three main approaches to performance monitoring. All three are based on different methods for collecting data and measuring response times:
- Monitoring from the end user perspective (a robot plays predetermined scenarios to evaluate general and user-specific transaction times.)
- Network and application performance monitoring (Application traffic across the network (between clients and front-end servers as well as within the application chain) is recorded across a number of measurement points.
- Agent-based application performance monitoring (agents on all servers in the application chain monitor execution of the application at program and system resource levels.)
Obviously, one would pursue all three approaches under ideal conditions (unlimited resources in the form of employees, time and money). However, since this is rarely the case, prioritization is necessary. Here you can see the advantages / disadvantages of the individual approaches:
|N/APM-Typ||End user view||Agent-based||Network Based|
|Questions answered:||• How has the Quality of Experience developed in relation to the standard procedures of my most important applications?||• Which link in the application chain causes the slowdown?
• Which transaction is slow?
• Which part of the program code has long durations?
|• Is there a slowdown?
• Which applications and users are affected?
• What causes the slowdown (network, server, data transfer)?
• For which transactions?
|Border:||• Not matched to actual usage / user behavior
• Not self-explanatory (does not tell you what or why something fails)
|• No view of the application availability in the network
• No view of the application environment (network, other applications ...)
|• no view of the cause of server-side problems (program code, server resources ...)|
|Limitations:||• Agent or robot must be deployed on each client
• Configuration of each scenario
|• Agent must be deployed on each server
• Impact on server performance
|• Traffic recording via SPAN or TAP|
|Scope:||• One or more critical applications||• One or more critical applications||• All applications distributed over the network|
|The main users:||• Helpdesk, reporting for management||• Development teams, quality assurance||• Infrastructure team, helpdesk|
Why is network and application performance monitoring more appropriate for most organizations?
We believe that among the various performance management solutions, the network-based approach is the most common approach for most companies. Here are the reasons why these solutions are better:
- Number of applications: Hospitals can not afford to implement an implementation for each application chain. It would be far too expensive for these multiple scenarios to implement server agents or to set up robots - both financially and personally.
- Network provision is indispensable for many applications. The network-based approach is the only one that takes this into account.
- Operating costs: Most companies can not afford immense financial and human capital investments for performance management. They need easy-to-implement and maintain solutions.
Selection criteria for your performance management solution
When you select a network and application performance monitoring solution, it is important to choose one that is specifically tailored to your needs.
|Number of applications:||A performance management solution has to deal with tens of thousands of different applications. You should reject approaches that are based on filtering applications to restrict performance measurements to a few critical metrics. Many applications are critical. Infrastructure teams do not always know in advance, which is critical for each department. It is indispensable that the solution collects performance metrics for all processes that are cut-to-size. In the same way, the analysis environment should be restricted to a set of predefined applications (HTTP, SQL).|
|IT resources required:||The solution should not require additional resources, but save your current resources time. It is very important that all administration aspects of the Performace Management solution remain as simple as possible:
Simplicity of installation
Simplicity of administration (updates, etc.), integration into your SNMP-
|Operating costs (TCO):||The cost of the application must be within the budget of company IT. You should consider the cost of extending the coverage through the performance management solution for your future architecture (e.g., multiple active data centers, virtualized)|
|Distributed solution:||The additional cost of providing the solution should be in one|
|Support for virtualization:||The solution should provide several options for collecting virtual traffic, for example, traffic between multiple virtual machines|