LPAR Monitoring and Reporting (AIX, pSeries)

Logical Partitions (LPARs) are robust micropartitions on IBM pSeries servers running AIX. Essentially, LPARs are virtual, independent systems running on a single server. Each LPAR can take advantage of a fixed amount of such system resources like CPU, memory, and input/output devices.

Using up.time's powerful LPAR monitoring, LPAR reporting, data collection, and graphing, you can optimize the micropartitions that are running on your IBM pSeries servers. Not only will you have a more efficient and easier to manage AIX and LPAR environments, but you'll also need to devote less of your IT budget to the hardware required to run your servers.

 

Consolidation

 

For example, you have a number of production servers running AIX and SuSE Linux that are not being used to their full capacity. Perhaps a server could be running one or two applications and not taking full advantage of  the hardware. Instead of wasting resources, you can consolidate these servers in a virtual, micropartitioned environment on an IBM pSeries server.

up.time's Server Virtualization Report can help you to pinpoint physical servers that can be combined on a single virtual server. The report highlights servers that do not fully use their CPU, memory, or disk resources. The report factors in CPU, memory, network, and disk I/O over a period of time and calculates an appropriate mix of workloads. Instead of wasting time with guesswork, you can efficiently virtualize servers in your network using actual data.

 

Monitoring Logical Partitions

 

up.time agents gather a deep set of metrics from the LPARs on an IBM pSeries servers. up.time's LPAR monitoring and LPAR reporting can collect workload information from LPARs that are managed by a Virtual I/O (VIO) on the pSeries server or ones that are on servers managed by IBM's Hardware Management Console (HMC).

When working with LPARs that are managed by a VIO, up.time collects workload data from the VIO and the individual LPARs on the system via an up.time agent.

 

 

If, on the other hand, an HMC is managing your LPARs, you would use up.time's Auto Discovery feature to detect the managed servers and add them to up.time. Working in conjunction with the HMC, up.time polls the up.time agents installed on the VIO and the LPARs for a specific set of workload data.

 

 

Reporting and Graphing

 

Once up.time has collected workload metrics from one or more LPARs, what do you do with all of that information? Use up.time's LPAR-specific reports and graphs to visualize the performance of each LPAR running on a pSeries server.

The LPAR CPU Utilization graph enables you to better determine the CPU entitlements (which indicate the amount of CPU power that is assigned to an individual LPAR on a system). You can use the graph to give you a clear view of how much you may need to increase or decrease an LPAR's CPU entitlement.

The LPAR Workload report gives you insight into the overall workload on an IBM pSeries server. This enables you to accurately adjust the CPU entitlements of the LPARs on a server and keep track of their overall workload over time.

 

 

Both the LPAR Workload report and the LPAR CPU Utilization graph take the guesswork out of determining CPU entitlements for the LPARs on a pSeries server. For example, your report or graph indicates that an LPAR with hard entitlement (one that cannot use spare processing power from another CPU on the server) has CPU usage which is constantly at or near the maximum. You can either increase the CPU entitlement change it to a soft entitlement (one which can use spare processing power from another CPU on the server). If, on the other hand, the LPAR has a soft entitlement and its CPU usage is consistently at or greater than the entitlement, you can increase it.

By using up.time's reports and graphs, you take the guesswork out of monitoring and optimizing your LPARs. You're using real data instead of relying on trial and error. Lern more about up.time's AIX monitoring here.

 

LPAR Performance Metrics for LPAR Monitoring and LPAR Reporting

 

The up.time agent that is installed on a logical partition of an AIX LPAR system collects the following performance metrics:

  • lpar_id

    The unique identifier for the logical partition (LPAR).

  • putt

    The amount of CPU time that has been used by the partition.

  • entitlement

    The CPU entitlement of the partition.

  • cpus

    The number of CPUs on which the LPAR can run.

  • capped

    Determines whether or not the entitlement of the LPAR is capped. If yes, then the LPAR cannot exceed its entitlement. If not, then the LPAR will be able to exceed its entitlement if there are spare CPU cycles available from another CPU on the system.

  • smt

    Determines whether or not hyperthreading is available.

  • memory_total

    The total amount of memory available to the LPAR, measured in bytes.

  • memory_used

    The total amount of memory being used by the LPAR, measured in bytes.

  • net_packets_tx

    The number of packets of data that have been sent over the network interface from the LPAR.

  • net_packets_rx

    The number of packets of data that have been received over the network interface by the LPAR.

  • net_bytes_tx

    The amount of data, measured in bytes, which has been sent over the network interface from the LPAR.

  • net_bytes_rx

    The amount of data, measured in bytes, which has been received over the network interface by the LPAR.

  • disk_kbytes_out

    The amount of data, in kilobytes, that has been written to the disk.

  • disk_kbytes_in

    The amount of data, in kilobytes, that has been read from the disk.