The follow-up of data-processing applications Quality of Service requires powerful tools and accurate competences. The produced and published data must make it possible to appreciate with accuracy the infrastructure Quality of Service total level and the real defects impact on users.

Performance Managers has developed a whole set of tools and competences to quickly implement an application monitoring platform, making it possible to measure data-processing applications Quality of Service, to validate measurements adequacy to the Contract of Service, to provide clear data for the defects resolution, and to even start preventive actions before real users do not feel the impacts of them.

Concepts

When a user sends a request to a server, the request passes initially through security and load balancing equipments, then the front-end server receives the request, then back office applications servers process it by calling data on RDBMS servers, and return back the answer to the user. Each component, each processing step can introduce delays or defects for the user. Whereas the majority of SNMP based supervision tools detect the components dysfunctions separately, one can evaluate the real impact on users only with real user emulation tools.

Performance Managers has developed a Quality of Service Monitoring platform, including installation of probes PCs emulating real users behaviour, to collect applications response times and availability, from anywhere on the network. Performance Managers installs several probes PCs to supervise simultaneously the infrastructure various segments.

Détails

Performance Managers sets up for you probes PCs at adequate network nodes, under exploitation staff agreement. A probe PC is generally a laptop unit, running Linux, on which runs Performance Managers load engine USERS, with a script simulating the real user behaviour. The target servers do not make any difference between a true user and the simulated user. The robot sends the script requests to the target architecture, according to the network protocol (WEB, Oracle, C/S, SNA3270, Tuxedo...)

The response times are measured from beginning to end, obtaining the full right answer from the server. One can obtain very precise details over the response times, in particular: connection time establishment, SSL encryption mode, DNS Time, each request time, total response time... The measurement interval and the measurements count are regulated according to defined needs for QoS measurement. Brought closer measurements make it possible to track the least defects. The segmentation makes it possible to precisely establish each infrastructure processing step contribution in the response time’s composition.

Performance Managers has a processing Center where correlation, presentation and reporting servers collect and compile measurements. The probe PCS send with regular interval the measurements to the Performance Managers processing Center. Results are sent as compressed emails of last measurements, through the customer’s SMTP gateway. There is no permanent connection of the probes to Performance Managers processing Center. Thus the supervision platform presents total safety.

The correlation server receives and stores the data resulting from measurements. The presentation server shows the results as real time graphs. The access is only reserved to known users. The reporting server sends formatted reports (csv, graphs, pdf), according to the selected periodicity (day, week, month), as emails to known users. Other transmission modes can be selected, like FTP, SNMP,…These data files can be automatically integrated to the customer’s supervision database tools.

Monitoring Plateform

All servers and Performance Managers probes use the Linux operating system. Remote support and updates are simplified. A probe PC can monitor up to 50 different applications, including with raised frequencies, and with a very high degree of measurements accuracy. The probe PC can carry out several simultaneous tasks (measurements, statistics, results sending..) without affecting measurements.

Implementation

All the implementation and support are ensured by Performance Managers. We define the measurements protocols, restitutions, then we carry out measurements scripts, we install and parameterize the probe PCs, we bring into service the correlation servers. Catch in hands by professionals, the Monitoring platform’s implementation is done as soon as possible, and one obtains reliable results quickly. Performance Managers intervenes like council throughout the service to explain results and to assist the customer, to arbitrate interpretations for infrastructure incidents resolution.

Results