We are constantly looking to improve performance, efficiency, and scalability of Foglight for Virtualization, Enterprise Edition (FVE) to help our customers get more out of FVE. As the ever growing number of Virtual Machines (VM) in IT organizations increases, there is an expectation that an instance of FVE will monitor even more virtual machines. In this blog I would like to highlight some of the improvements that are in the FVE 7.2 release, in which we have measured significant improvements in the product performance, efficiency, and scalability. Note: FVE 7.2 is scheduled to be released in Aug/2014.
1) Optimized the data submission between Agent and Server
- Instead of submitting all data collected in a single large submission, the data collection is improved by allowing many smaller submissions per data collection. This allows the server to control the rate at which it receives submissions and assign multiple threads to process them.
- Elimination of duplicate values submitted by performance collection, for NICs and Disks
- Enhanced data collection quality by improvements in data collection intervals, sample rate, and number of collections
Default FVE collection
- Data collection interval: every 5 minutes
- Sample Rate: every 20 seconds
- Number of collection: 15
2) Consolidation of frequently executed Rules/Derivations into a single batch execution. Here are some of the consolidated items:
- Swap In/Out Rate Metrics
- Growth Rate Derived Metric
- Uptime Percent Derived Metric
- VM Baloon Memory Rule
- Fill Time Rule
- Free Memory Derive metric
3) Decreased the number and frequency of Database calls from the Server
- Improvements in batch query rendering the Virtual Infrastructure Overview Reports. This has resulted in significant time reduction in generating the reports.
- Batch Query Optimization – Restrict Batch Queries to 100 Objects per query to ensure VM Explorer UI is populated accurately and quickly, and also provides a more normalized DB load
- Improve performance of Virtual Machines tile in large VMware environments
- Limit vCenter Batch queries to VMware recommended range, and make it configurable
4) Reduced or eliminated topology churn for many topology objects
- Missing VMware Objects – Improve Service Tree rebuild using the latest topology object version to ensure the most recent changes.
- Duplicate VMs - Corrected duplicate name and UUID fields in ESXServerNode during performance submission.
- Topology changes – Dependencies refresh when the vCenter object has topology change
5) Improvements in data refresh and data consistency
- Display the correct information for ESX Hosts, Cluster, Datacenters, and vCenter when the object(s) are removed from the Agent Black List
- Calculate and display the correct Perspective Tile in Exchange Mailbox count
6) Enhanced metrics and derived metrics in various areas
- Enhancements to FreeMemory derived metric in Capacity Manager
- Corrected the CPU load presented for a vAPP. It is now collected based on shared and consumed CPU
- Improved the value calculated for Datastore Cluster Latency by using the weighted latency values for the sub-datastores
- Improved the algorithm to calculate workload in Capacity Manager
7) Improvements for Product Upgrades
- The Foglight Installer is enhanced to detect custom install paths during upgrade
- Enhanced Foglight upgrade to handle dependencies during upgrade
Our goal is to deliver incremental improvements in FVE performance, efficiency, and scalability with every release. In the 7.2 release we have allocated significant efforts towards this goal which have resulted in substantial improvements. For the complete list of all improvements in FVE 7.2 release please refer to FVE 7.2 Release notes.
Foglight for Virtualization Enterprise Web page: