PlateSpin PowerRecon helps plan for VM growth
- 12 February, 2008 08:05
PlateSpin's PowerRecon is a planning and monitoring application for organizations with a high number of servers and virtual host targets. On the surface, PowerRecon looks similar to traditional network monitoring/management applications that track application inventory, connectivity and network usage.
We like it, however, for what's underneath, which is a very useful server monitoring, planning and management tool (especially for VM hosts and instances) that can be coupled with PlateSpin's PowerConvert tool, which does a formidable job of moving server instances from physical to virtual and all the variations the virtual data center affords us today.
PowerRecon also has chargeback features that monitor and levy costs based on actual system uptime, CPU utilization, processor speed, memory used, network bandwidth used, and disk storage (space, speed and number of writes per second). By contrast, Virtugo's virtualSuite uses a patented, contrived metric for chargeback accounting purposes.
We were also impressed with PowerRecon's ability to work with a huge number of VM operating system candidates, including Windows NT 4 and XP through current Windows Server products (support for XP and Win 2003 Enterprise Server were tested and affirmed), most modern versions of Red Hat and SUSE Linux (Red Hat Enterprise 5 was tested) and Sun's Solaris 7 (Sparc) and Solaris versions 8-10 (Sparc and Intel/AMD). VM hosts can be running VMware ESX (tested), Citrix Systems' XenServer Enterprise Edition, Virtual Iron and/or Microsoft's Virtual Server 2005.
PowerRecon and associated daemon modules run as a Windows service application. Once installed, PowerRecon uses a Windows Explorer-like user interface called Data Center Explorer that performs a network discovery and maintains an inventory of hosts. Credentials are then supplied by the administrator for server hosts that need to be internally inventoried or further monitored. A handy list of credentials are provided but placing so many crucial passwords in a single place, even though they are protected by files and database access permissions, made us nervous.
The results of the inventory process were shown in a tabbed user interface containing descriptions of host or VM instance characteristics -- such as processors, general information, disk drive information, network adapters, applications, operating system services/daemons, and any other user-defined fields. Power dissipation is also tracked.
Each host can be assigned to a group or site, and we could easily start/stop monitoring, force an inventory to be taken, or launch an application, like a terminal services/remote desktop connection, an SSH or VMware Remote Console session, or another management application such as Embotics' V-Commander. A new template for tracking purposes could also be launched that allowed us to track operational fundamentals of the VMware ESX host we tested with, or a VM instance we tracked. We could set the thresholds for tracking and viewing graphical data, such as CPU utilization, memory use, disk space, disk I/O, and network I/O for warning purposes and watch the trendline or use collected data to graph a forecasted trend.
Once fully configured, PowerRecon can inventory, or begin reporting monitoring details on demand or on schedule, and report findings via e-mail (an available separate SMTP server is required) or to various importable file formats. We set and forgot about this feature until reports started showing up like clockwork. Reports are highly configurable.
Action from information
PowerRecon has a five-step server consolidation routine that divides potential mergers into either server and/or workload protection projects. We selected a server consolidation project based on existing discovered inventory, then gave PowerRecon a date and time range (varies for an hourly or user-specified time period that can exclude specific work days like Saturdays and/or Sundays) then had PowerRecon forecast/project a load with an optional power/cooling factor. Trend data flushes after 30 days, which can be a limitation. The data is then gathered and we examined it to determine which consolidation/protection candidates existed among our test server groups. It turns out that we don't utilize our servers enough and most of them were candidates.
Over the term of using PowerRecon, we came to depend on it as well as Onaro's tools for our every-day lab and production activity sources. PowerRecon doesn't have quite the luscious user interface of SANscreen VMInsight, but is very useful nonetheless. Coupled to optional PowerConvert, PowerRecon is a strong combination that's useful for medium to large installations.