United States (change)
Shortcuts: Downloads Fedora Red Hat Network
Red Hat Command Center offers monitoring software as a service, hosted by Red Hat. In the diagram below, the Command Center application, represented by the box on the right, is hosted and delivered from Red Hat's datacenters. This backend is based on a mature technology that was designed with a multi-tenant architecture from the ground up, instead of layering software as a service capabilities on a standalone application. As a result, Command Center maintains tight security and privacy for each customer, while passing on the cost saving benefits achieved through economies of scale. Moreover, the application is deployed over an n-tier, farm-based architecture, allowing it to be highly scalable across multiple systems, applications, networks, or locations.
Command Center requires the deployment of a scout locally on the customer network. The scout is essentially a go-between, used to collect the monitoring information. The scout runs a variety of probes against your monitored hosts, using common application protocols such as SNMP or HTTP to gather monitoring data. This data is then sent to the Command Center backend over a secure connection. The backend handles alert notifications and provides monitoring status and reports through the web portal.
Red Hat Command Center is simple to set up and use, requiring no specialized skills. Typical steps include:
The scout is not the Command Center application, but merely a conduit to collect monitoring data. If the scout hardware were to die, it can be easily replaced with another scout. After a handshake between the new scout and the customer account, the monitoring solution would be restored with the customer configurations, schedules, and historical data. This is because the Command Center backend is the core monitoring application, where most of the processing for the incoming data, monitoring analytics, and data storage takes place.
The scout allows monitoring of behind-the-firewall IT infrastructure without the need to open incoming network ports. It is a locked-down appliance that is configured with no listening ports. All scout communication to the target monitored systems, as well as from the scout to the hosted Command Center backend is one-way and initiated by the scout. The scout collects monitoring data from the customer environment at the polling frequency defined for each probe. The scout packages this data and sends it to the Command Center backend at regular intervals over HTTPS. The scout also requests any updates to the configurations, probe definitions, etc., during this dial-home connection session. The Command Center backend seamlessly delivers any changes that the customer may have made via the portal, as well as new feature or bug fixes available after a product release via this mechanism. This allows the scout to be maintenance-free for the customer.
The scout appliance is based on Red Hat Enterprise Linux 4.5 and requires standard x86 hardware that is compatible with this distribution. The minimum system requirements include 1.6GHz CPU, 512MB memory, 10GB disk, and a boot-capable CD-ROM drive. This configuration can easily support up to 200 servers. For larger environments, more memory is certainly recommended. It is also possible to distribute the load across multiple scouts and consolidate the monitoring through the unified dashboard view.
Command Center is based on the philosophy that monitoring should be be as non-intrusive as possible. In most cases, the Command Center scout uses common application protocols such as SNMP, HTTP, etc., to collect monitoring data, requiring no agents or changes to the target system. In other words, Command Center monitoring is agent-less for several types of probes, such as host up or down probes; network services probes such as DNS, FTP, etc.; application level probes including JBoss, Apache, Tomcat, Oracle, MySQL, Microsoft SQL, etc.; as well as device and website monitoring probes.
For other types of probes, such as CPU or disk utilization, a lightweight SSH-based daemon (or WMI service on Windows) must be installed on the target system. These "agents" have a very small footprint. When requested by the scout, the daemon invokes native commands on the target machine (such as "df" for disk usage on Linux/UNIX), returns the data, and goes back to sleep. The daemon imposes almost no load on the target machine, as it does not run any local processing logic, nor does it store anything locally. The SSH daemon uses an unprivileged account and accepts only encrypted communications from the scout. This ensures that the security on the target system is not compromised.
Systems that require a lightweight daemon are marked with * in the Command Center probes sheet (pdf).
The scout packs the data from various probes and incrementally sends it to the Command Center backend. Even for a deployment with more than 200 servers, the data exchanged between the scout and the Command Center backend every few minutes is no bigger than a typical web page, like the one you are viewing right now, and can be easily transmitted over a low bandwidth connection.