The analysis function enables operatives and people in charge to analyze how a data-center works, be it on a daily basis or on the long run. This feature includes both a performance board and a tracking table.

The performance board

The board provides a real-time status of the data-center performance for operational teams. It aims to reveal possible ways of performance improvement for the data center. The objective is to give a synthetic global picture to quicken decisions for corrective actions. An alert system can also be set up by the administrator. The main key performance indicators are currently CPU, RAM, drive access, network load and electricity consumption.

The tracking table

This table has been developed for the team leaders and shows the performance changes of the data-center over time. It uses acknowledgements to help the supervisory staff understand what happened and to communicate about successes and failures. The table performs surveillance. In that table, the btrCloud solution enables over- and underloaded servers and VMs to be detected, as well as disabled and orphaned VMs. This solution also enables more accurate infrastructure analysis, such as measuring the shared memory rate on a specific physical server, or identifying the servers which drive access resources are damaged, by locating the bottleneck(s) in the infrastructure.

btrMonitor is used for reflectively analyzing the infrastructure, whereas btrAction performs intercession of the physical and virtual worlds. Both of these modules use a system with autonomous drivers for each addressable hardware and hypervisor. This results in a separation between the module itself with its functionalities, and each driver contained in an independent Java ARchive.

btrMonitor Monitoring consumptions in the computer stock

Concretely, btrMonitor periodically examines specific hardware (hypervisors, management cards, wattmeters…) to collect the available consumption data on the infrastructure by means of the embedded drivers.
If a server electricity consumption is unavailable, btrMonitor will then estimate it as a function of the server CPU consumption and the maximum (with 100% CPU load) and minimum (with 0% CPU load) server power consumptions, provided that this information is available in the database.
The time periods between two data measurements or between two data sending to btrPersist and btrScript are configurable.
Besides, the flexibility provided by the driver-based architecture and TCP communications allows parallel deployment − ideally on different servers and/or VM − of multiple instances of btrMonitor that manage some network subparts. This is also true for btrAction.

btrAction Logical restructuration of the data-center

btrAction enables the computer stock to be driven, regardless of how performed actions are triggered. There are two possibilities:

The aim is to enable shutdown, standby and restart of the servers and VM; and more specifically for the VM: migration from a server to another, snapshot creation, physical destruction, etc.
Of course, all of these actions are not available with each driver. Typically, a management card driver does not allow the VMs of the server which is managed by this card to be managed themselves. Besides, multiple drivers may be used to manage one single server.
From the user side, the management is transparent: first, btrAction knows which driver to use to perform a given action on a given (physical or virtual) machine, and an error message is sent for each action that would not be compatible with the available drivers on the server.

btrView and btrConsole enable administrators to introspect and act on the data center. btrView is a GUI administration console which provides an overview of the information about the cloud. All the btrScript actions are available through this console, which also has a configurable alert system. The btrConsole module is a textual console which enables all the functionalities given by btrScript to be used.

btrView A lightweight client GUI

View of a server in btrView

View of a server in btrView

1: Tree

The tree describes the entire cluster, with icons of different colors depending on the state of VM (red for “Stop”, orange for “Suspend” and green for “Run”).
With this tree you can do all actions on machines, using the context menu (to change the state or apply rules) or using the drag & drop for migrations (this is also possible with the right click in the context menu).

2: Data analysis

This section displays the graphs of various data in real time. There is the consumption of the CPU, of the RAM, and of the energy in graphs. When you select a physical machine, there is also an overview of its virtual machines, and a graph of the consumption in watt.hour. In both cases (with a physical or a virtual machine), you can also see the list of effective rules on the selected machine.

You can select the cluster too, to see its total consumption of energy in watt and in watt.hour.

Two graphs are also proposed in the tabs of the analysis: the sankey diagram showing the distribution of CPU, RAM and power in the data center, and treemap to visualize virtual machines of data center in blocks.

3: Other features

Analyze The button “Analyze” is the previous point, the graphs overview.
Rules The button “Rules” is the list of all the rules in the cluster, with possibility of deactivate/activate/delete rules. Rules can be created by right-clicking on a selection of several machines in the tree.

Schedule The button “Schedule” is where you can plan actions or rules that run when asked.
Scheduling's screen

Three tabs to create: one for actions, one for the rules, and another for more freedom.

Errors The button “Errors” shows the conflicts in placement rules.
Conflicts' screen

Here, the rule that vm.s3.9 should be on is in conflict with reality: it is indeed off.
These conflicts are also displayed in the logs.

btrScript The button “btrScript Console” is a command-line interface for direct dialogue with btrScript to obtain information not available in btrView, or to execute some actions in command lines.
Consoles' screen

In the console, the administrator can query directly btrScript.

5: Current actions / Logs

You can see current actions (ex. “vm1 is starting”) with a progress bar, and logs of actions, failures or errors.

6 : Energy consumption

This zone is present to show quickly how much the data center consumes energy, and the energy that could be saved with the btrCloud solution.

btrConsole A textual console

Coming soon…