This article is part of an Essential Guide, our editor-selected collection of our best articles, videos and other content on this topic. Explore more in this guide:
3. - Using performance metrics to enhance storage for virtual environments: Read more in this section
- Performance monitoring for virtual server storage takes application approach
- Using virtualization-aware storage for per-VM statistics
- Storage for virtual environments needs better performance monitoring
Explore other sections in this guide:
- 1. - Managing storage for virtual environments: Best practices
- 2. - Top tools for virtual server storage management
- 4. - Hypervisor-specific data storage management
As workloads grow more complex and become virtualized or are moved to cloud environments, IT professionals can no longer measure infrastructure performance with gut instinct and basic tools. As a result, server performance monitoring tools must evolve to holistically monitor and analyze data center operation.
Data center infrastructure performance monitoring lets IT professionals keep a watchful eye on the performance of servers, as well as other critical infrastructure components. Monitoring tools are increasingly necessary to manage virtualized resources, which can change at a moment's notice. IT professionals proactively shift workloads based on resource use and performance data, and can even anticipate changes in workload demand and plan capacity accordingly.
"Everything we have is virtualized or will be, from our SQL database and related time and billing applications to our Exchange server, which is slated for migration to a virtualized environment soon," said Michael Ferguson, the director of IT for Miami law firm Rennert Vogel Mandler & Rodriguez P.A.
Ferguson's production servers are in a colocated bunker and also in a separate location in Boca Raton, Fla. "My daily concern is keeping consistent bandwidth between Miami, where we have 60 people, and our server in Boca Raton. We are always tweaking and watching -- and fortunately we have high availability in both VMware and storage virtualization in the form of SANsymphony from DataCore," he said.
In that new virtualized world, monitoring is a challenge. "We have spikes and valleys ... in the flow of what is happening at that particular moment," Ferguson said. "For instance, if the real estate department is handling a lot of closings and paralegals are executing a lot of documents within the software, we see spikes. But that is only occasionally." Future performance is also an area of concern. "I have to anticipate things like how [VMware] vCenter running ESX servers is performing. Last week, the reports indicated I needed to increase RAM, so I have ordered more," he said.
Monitoring capabilities built into VMware and SANsymphony are currently his mainstays, because his biggest concern is running short on storage. Ferguson said while his existing management tools are adequate, he would like a single tool that could provide a comprehensive overview of operations and performance.
The market is saturated with server performance monitoring tools, O'Donnell said. The key players are the usual suspects in management: Hewett-Packard, CA Technologies and BMC Software, and there are even a number of open source server monitoring options, particularly Nagios, which is getting very popular but "suffers from complexity," said Glenn O'Donnell, Forrester analyst.
Monitoring on critical production systems should be pervasive, according to Jonah Kowall, Gartner research director for IT operations management, in the report "How to Augment APM with Conventional Monitoring." Beyond server performance monitoring, tools should focus on application performance management (APM) and monitoring, since users ultimately rely on application performance.
A combination of server, network, database and storage monitoring, along with more in-depth app performance monitoring, can deliver a complete view of the application and its performance, Kowall said.
Analytics in the infrastructure monitoring toolset
Advanced algorithms can "help analyze the mess on the back end, including understanding behaviors and why things are happening," added Glenn O'Donnell, Forrester analyst. This is how application performance monitoring and other kinds of monitoring are coming together, he said. "APM tells you that you have an issue and, in theory, the analytics will tell you why and will help you drill down and map the problem."
More on next-gen monitoring
Learn about application performance monitoring in part two of this series.
However, he cautions, analytic capability for monitoring is still in its infancy.
The key is encouraging IT professionals to acquire more sophisticated analytics capability as they can get it. O'Donnell said there are a lot of useful management features coming from big companies, such as IBM with its Tivoli management software, but most of the innovation is still coming from little "peanut companies." Over the next few years, the further development of analytics will make monitoring much more exciting, he said.
Ferguson finds that kind of broad capability -- to supplement what's built into his SANsymphony storage manager -- very desirable. "The second thing on my wish list after 'Don't bust the budget' is a dashboard app that would reach into the infrastructure and provide at-a-glance reports I can drill down into," he said.
If a utility included application performance, that would be a plus, Ferguson said. The billing, accounting and financial management software application at his firm "gets used constantly, but not with a big hammer." Application performance monitoring would be more welcome for the firm's most demanding applications -- built on SQL Server and Linux-based MySQL -- especially during end-of-month closing activities.
"If you have a good dashboard, things go well. The instruments on your panel, if you trust them, can get you a long way," he said.