- Capacity and real-time performance management for 2,000 heterogeneous physical and virtual servers
- Optimization of audit services to improve performance management
- 60% reduction in audit time
- Shared capacity management repository
- Application managers involved in performance management
- Internal chargeback of audit services
- Preparation for private cloud computing
The deployment of Axway OmniVision helped us prepare the ground for better control over new architectures, especially in virtualized environments, which will further our efforts to implement an internal cloud.
EDF France is part of the global EDF Group that provides tens of millions of businesses and individuals across France with electricity. While seeking to reduce IT costs through the benefits of infrastructure virtualization, the company’s IT leaders knew they needed to better manage capacity and improve performance across their 2,000 production servers. They turned to Axway OmniVision.
“The deployment of OmniVision helped us prepare the ground for better control over new architectures, especially in virtualized environments, which will further our efforts to implement an internal cloud,” said Charles Le Jossec, IT Operations Deputy Director at EDF.
To start, the EDF Operations Division tackled optimization of processing capacity of its server farm that was running various versions of AIX, Windows and Linux. “As CPU usage was only at 20%, we applied an aggressive consolidation policy based on micro-partitioning and virtualization. To run this strategy, part of an ITIL-based standardization approach, we selected OmniVision to manage the availability and performance of our server resources,” Le Jossec explained. “Since implementation, we have had a single tool for optimizing the performance levels and adaptability of our three targeted architectures.”
OmniVision reports enabled EDF to identify applications suitable for virtualization and to determine the size of the targeted machines. In the production phase, these same metrics allowed the company to manage overbooking on AIX servers. As a result, CPU usage is now over 50%.
An Objective Technical Interface
Facilitating communication and optimizing performance
The data collected by OmniVision is also used as an objective technical interface that facilitates communication with the company’s three outsourcers as well as internal customers. The commitments and requirements of all parties are evaluated using a common framework. The portal centralizing the metrics collected by OmniVision is consulted on average 600 times per month. This indicates that OmniVision is properly fulfilling its role as "peacemaker."
"Currently, some 2,000 production servers are equipped with OmniVision, 50% of which are virtualized. After helping us establish and manage availability of a dynamic infrastructure, the software is now helping us optimize performance. It was the first step towards cloud-based management of our internal resources," Le Jossec said.
Performance and capacity management are the keys to getting the most out of heterogeneous and virtualized environments.
The Transformation of EDF’s Information System
Since 2004, EDF’s server infrastructure has increased by 80%, totaling some 6,000 machines, including 2,000 in production. Over the same period, the number of applications has increased by 140%, and data storage has grown by 980%.
The scale of these metrics demonstrates the IS transformation at EDF, which, in just a few years, has transitioned from being a state-run utility to a dynamic, public company.
To optimize performance management services, EDF also deployed OmniVision Investigation. Using the same OmniVision agent already installed to manage capacity, this real-time system parameters collection and analysis function enables Ghislaine Marchand, Head of Production Control, to focus the audit department’s efforts on more value-added activities.
Due to the lack of an appropriate measuring tool, the audit department was previously spending 60% of its time installing scripts on each server and retrieving, formatting and interpreting performance metrics. Despite best efforts, however, the department was unable to attain satisfactory levels of quality and responsiveness, particularly when it came to crisis situations that required quick resolution.
“OmniVision Investigation improved reliability and entirely automated the collection and formatting tasks for 500 metrics across a heterogeneous production infrastructure,” explained Marchand. Furthermore, the flexibility of the OmniVision portal has greatly simplified the configuration of customized dashboards. “Giving application managers direct access to selected metrics involves them in managing the performance of their applications and makes them aware of capacity management.”
OmniVision improves reliability and entirely automates the collection and formatting tasks.
Comprehensiveness and Granularity
Freed from tasks that ultimately offered little value, the audit team now concentrates on managing capacity and performance. “Thanks to the comprehensiveness and granularity of the metrics collected by OmniVision, we can now offer more complete reports that include measurements across the entire heterogeneous server infrastructure,” said Marchand.
The OmniVision database and portal enables users to customize and manipulate combinations of data collected and measure the availability of resources for a virtual machine, a server, an application component, a consolidated application or even as a business. Both the frequency of metrics collection (from per second to per hour) and their retention duration can be adjusted. “From this repository, we can reproduce an incident in full detail to identify its root cause.”
The resources freed up by automating access to performance and monitoring data from load testing can instead be invested in setting up a genuine performance audit service. “We are meeting our deadlines and producing more relevant reports,” concluded Marchand.