|
Proyecto BPM
| |
CERN |
| |
|
| |
CHILE |
Sector AAPP |
| |
|
|
| |
DESCRIPCION |
RETO |
|
When you’re trying to understand the building blocks of matter, the origin of the universe and everything in between, system downtime is not an option. CERN, the European Organization for Nuclear Research, explores these complex, epic issues using the world’s largest and most powerful particle accelerator, the Large Hadron Collider (LHC).
CERN’s goal is to recreate the Big Bang with the LHC, using its thousand 15-meter magnets to fire particles into each other. With particles colliding at the speed of light, the Technical Infrastructure Monitoring (TIM) system tracking LHC operations must be reliable. That’s why CERN adopted Progress® SonicMQ®, the industry’s most robust and resilient enterprise messaging system, as the communications infrastructure for TIM.
|
Located west of Lake Geneva, CERN is spread out over Swiss and French territory. The most important systems, such as the Large Hadron Collider, are underground. The LHC is in a circular tunnel with a circumference of 27 km, the latest addition to CERN’s existing linear and circular accelerators.
In the LHC tunnel, more than 1,000 super conducting magnets, each 15 meters long, will accelerate elementary particles—protons or heavy ions—through a vacuum filled tunnel to 99.99% of the speed of light. When these particles are fired into each other in opposing trajectories, their impacts involve extremely high energy levels. Physical conditions similar to those immediately after the Big Bang are created for fractions of a second.
It’s no surprise that it takes a highly complex technical infrastructure to run this huge research laboratory and its scientific systems. These includes the power supply—the LHC needs as much electricity as all of the households in the city of Geneva combined—ventilation of underground systems, refrigeration for the super conducting magnets, high-performance vacuum pumps, water pumps, air conditioners, fire alarms, communication systems and access control.
Currently, over 100 technical infrastructure systems with about 25,000 individual measuring points generate around 1.3 million individual items of information daily. When the LHC becomes fully operational, over 150 systems with 50,000 measuring points will be delivering around 2.6 million values a day.
This infrastructure is monitored round the clock by the CERN Control Centre (CCC), built for the LHC on the French CERN site in Prévessin. The data to be monitored is collected, evaluated, stored and distributed using CERN’s Technical Infrastructure Monitoring system (TIM). Using TIM, the status of any equipment—any vacuum pump, relay or valve—can be checked online at any time. This infrastructure information is kept completely separate from “physics data” and experimental results, which are managed in dedicated IT systems in a separate control center.
|
| |
|
RESULTADO |
| |
A high-performing, continuously available messaging system.
The information provided by the infrastructure systems is merged in TIM and provided to users via a graphical user interface, ensuring a consistent look and feel. A logging system integrated in TIM automatically stores selected data for further evaluation and subsequent checking. This is useful for post-mortem analysis of severe equipment malfunctions.
Just as important, TIM automatically informs the central LHC Alarm Service (LASER) of problems with the technical infrastructure that may be of interest to the different site services. In total, TIM contains about 20,0000 definitions of potential fault situations.
“TIM collects data from a broad range of different, widely distributed systems and also forwards control instructions to those systems,” explains Jan Stowisek, a CERN engineer responsible for the TIM system. “We have therefore implemented a flexible data acquisition mechanism that supports a large number of protocols and can be expanded by additional protocols if required.”
CERN started implementing TIM in 2003. It used J2EE because a number of systems at CERN were already based on this technology. Using similar hardware and software platforms for multiple systems significantly reduced support costs.
For its communication infrastructure, TIM uses Progress Software’s messaging middleware, SonicMQ, based on JMS. As an asynchronous, messaging-oriented solution, JMS is much better suited for the challenges TIM faces than a connection-oriented client-server approach such as those supported by RMI or CORBA. It is the only communication protocol between the data acquisition, processing and distribution layers.
“Also, JMS allows a clear separation of these layers, which means that the implementation of one layer is independent of the implementation of the other system components,” explains Jan Stowisek. “This is very convenient for maintenance work. For instance, we can perform maintenance on the application server without having to restart any of the client applications.” JMS is also used for communication with the LASER alarm system.
|
Real-time insight into Large Hadron Collider operations; zero downtime for the monitoring infrastructure; over 150 systems monitored with 50,000 measuring points delivering nearly 2.6 million values a day.
The initial TIM prototype was based on OpenJMS as a messaging middleware and JBoss as an application server. However, because a large volume of data needs to be processed, CERN required a highly available and high-performance solution.
Considering the significant investments made at CERN, failures of the control systems could have serious consequences and result in a substantial financial burden. Unplanned downtime is unacceptable for such costly systems, considering that physicists from all over the world who come to CERN for only a few days would be interrupted in their experiments.
“We tested SonicMQ and were convinced by its availability features, especially with regard to clustering and load balancing, and its flexible configuration options,” explained CERN Project Manager Anna Suwalska. “We also received excellent support by Progress Software staff during our tests and during our subsequent design work for the solution.”
In its operational configuration, TIM currently uses a cluster of two SonicMQ brokers to ensure optimal availability. A third broker installed on a separate server manages the cluster. On the hardware side, the TIM application servers and JMS brokers are hosted on HP Proliant servers. An additional server is standing by to take over immediately should one of the primary servers fail. The scalability of the SonicMQ messaging solution is also critical for TIM.
Currently, only the smaller CERN accelerators are being used for experiments. However, once the LHC is ready for operation, the amount of data to be processed will be far greater. For this reason, further servers can be added to the TIM infrastructure when necessary. SonicMQ now forms the backbone of TIM and, thus, of a large part of the technical infrastructure. A wide variety of information comes together on this bus and is distributed to various client systems.
“The solution is very reliable,” explained CCC Technical Infrastructure Operations Manager Peter Sollander. “Our success with TIM and LASER shows that J2EE technology is not just for eCommerce applications, but that it is also suitable for industrial process control,” summarized Jan Stowisek.
|
| |
|
| |
|
Fuente: Progress Software
|
|
