In the Performance Management Topic at this site we separate performance and capacity monitoring for virtualization and the cloud into four separate disciplines:
- Configuration Management – while configuration management is not traditionally associated with resource and performance monitoring, it is critical in the case in virtualized and cloud based environments. The rate of change in these environments and the degree of resource sharing of these environments makes a continuous self-discovered understanding of configuration essential to the ability to assure infrastructure and applications performance.
- Resource and Availability Management â€“ which focuses upon monitoring the utilization of the key physical and virtual resources that provide the ability to support workloads in virtual and cloud based environments. These resources include virtual and physical instances of CPU, memory, network I/O and storage I/O. These resources need to be monitored in near real time for performance management purposes, and trended over time for the purpose of capacity planning and management.
- Infrastructure Performance Management â€“ which focuses upon the response time of the infrastructure to requests for work placed upon it by workloads. This is an important and separate category from the above since it is our opinion that in virtualized and cloud based environments one can no longer infer the performance of an infrastructure from it resource utilization (for a variety of reasons covered in the Virtualization Performance and Capacity Management White Paper). Therefore a new category of tools are needed which focus upon measuring the response time of all or portions of the infrastructure on behalf of its supported applications. This new category essential to allow for the virtualization of Tier 1 performance critical applications.
- Applications Performance Management â€“ which focuses upon discovering the topology of an applications system (and keeping that topology map up to date in a dynamic private/public cloud environment), and measuring the round trip and per hop response times for each application running on the virtual or cloud based infrastructure.
- Service Assurance – an emerging category that will combine elements of all of the above disciplines with configuration management and the ability to guarantee the performance of the most important applications by measuring their response times, and then automatically allocating the correct levels of resources to these applications (any denying them to less critical workloads).
There are many vendors who are targeting the virtualization performance management and monitoring spaces with interesting solutions. The table below categorizes the vendors into the four main areas discussed above.
While the above table lists a large number of vendors, it also groups together vendors that have very different features and capabilities in their solutions. A brief review of each vendor and what they actually do is therefore useful.
Akorri invented and leads theÂ InfrastructureÂ Performance Management category for VMware based virtual environments. Akorri is the only vendor that provides a round trip Infrastructure Response Time metric from the VM that is interacting with the storage layer to the spindle on the disk and back again. This IPM metric is synthesized with an understanding of capacity utilization, which in turn is used to create a Performance Index that shows theÂ trade-offÂ between IRT and capacity utilization.
AppDynamics is one of the three creative applications performance management vendors who are run by former Wily (who invented and lead the J2EE APM market) executives. AppDynamics targets Java based applications running in internally hosted or cloud based environments. The product has two unique capabilities; the ability to map the topology of the application system, and the ability to work when an application is spread across public clouds and internal data centers.
AppFirst is one of two vendors (BlueStripe is the other) that can discover the topology of an applications system and provide round trip and hop by hop response times irrespective of how the application isÂ construedÂ andÂ irrespectiveÂ of whether it runs on Windows or Linux. AppFirst is also one of the few vendors that offers its product on a vendor hosted based (monitoring as a service or MaaS). AppFirst is also focused upon how the details of infrastructure resource utilization impact applications performance.
BlueStripe is one of two vendors (AppFirst is the other) that can discover the topology of an applications system and provide round trip and hop by hop response times irrespective of how the application isÂ construedÂ andÂ irrespectiveÂ of whether it runs on Windows or Linux. BlueStripe is offered as on premise software, and also supports applications systems that span physical and virtual infrastructures. BlueStripe also recently announced integration with Microsoft SCOM.
CA Technologies is on the only one of the big four “legacy” systems management vendors (IBM, HP, and BMC are the others) that has fielded an effective set of offerings for the virtualization and cloud computing markets. At this point CA has solutions in the Infrastructure Performance Management, Automation, Backup and Security areas. The IPM solution leverages assets acquired from NetQos, which was a leader in understanding the response time of applications from a physical and virtual network perspective. CA has yet to integrate its configuration management solution (CA Virtual Configuration) with its IPM solution, nor has CA updated the Wily APM solution for virtualization and the cloud. But CA is rapidly putting the pieces together, and integration is certain to follow over time.
Hyper9 offers the most flexible and comprehensive search based discovery and data collection for a VMwareÂ environment. Hyper9 includes deep and broad discovery of the environment and its configuration with collection of resource utilization based statistics on VM and host performance. This synthesized set of data can be effectively used to control sprawl, plan for capacity, ensure the integrity of the environment, troubleshoot the cause of performance issues and provide for business management of the virtual envionment.
ManageIQ is a broad scale virtualization management vendor, who also collects performance statistics from VMware Virtual Center. ManageIQ provides full management of the lifecyle of a VM, sprawl control, configuration management, the ability to provide a self service portal to business users with an understanding of resource utilization in the environment. ManageIQ is therefore not focused upon performance and capacity management as it primary value to the customer, but collects some of the same data that vendors that are focused upon these disciplines collect.
Netuitive collects the same vCenter API data that everyone else collects and then does something completely unique with it. Netuitive has the market leading dynamic data analysis and correlation engine applied to the problem of automatically finding performance and resource utilization anomalies in the virtualÂ environment. The use of statistical self-learning technology lets Netuitive be the only vendor that can provide accurate performance management of a virtual environment with a solution that requires no initial setup for thresholds, nor any manual maintenance.
New Relic invented and leads the business of providing applications performance management on a SaaS basis for Java and Ruby-on-Rails applications hosted in internal and public cloud environments with over 5,000 customers. Founded by Lew Cirne, the former founder of Wily, New Relic is turning the Wily model of APM upside down. New Relic offers a free basic version of its service, and offers the for charge versions based for affordableÂ monthlyÂ subscriptions. The product works out of the box with none of the expensive and laborious configuration that legacy on premise APM solutions require.
Platform Computing leads the market for managing resource allocations in the high performance computing market. Platform has brought these assets to a virtualization and cloud focused offering, Platform ISF. Platform ISF is one of two products available today in the Service Assurance space which is focused upon being able to guarantee the right resources to business critical applications so that performance levels are achieved.
Reflex Systems was the first vendor to leverage the vmSafe.net API, which provides Reflex a uniquely deep and real time perspective on configuration state and state changes in the VMware vSphere environment. Reflex is also a security vendor with deep packet inspection technology and can thereforeÂ identifyÂ applications on the network at a higher level of granularity than port and protocol. Reflex also collects the VMware vCenter resource utilization data. Reflex is therefore unique in the respect that it has great applicationsÂ visiblyÂ and great configurationÂ visibilityÂ combined with basic resource utilization statistics. Reflex is not a primarily a performance management vendor, but since Reflex has two key features of a true solution is this area, Reflex in conjunction with a true IPM solution might be compelling choice.
Veeam offers Veeam Monitor and Veeam Business View. Monitor collects the vCenter resource utilization statistics and provides reporting, alerting and capacity planning functions around these metrics. Veeam Business View allows these metrics to be presented organized in a manner relevant to business constituents.
Virtual Instruments is a leader in the Infrastructure Performance Management space with a focus upon real time and continuous instrumentation of the Fiber Channel SAN. Since issues with storage latency show up in SAN exchange completion times, this solution is an excellent way to monitor and ensure that issues with storage latency are not creating systems and applications performance problems. VI also collects the VMware vCenter Â and cross correlates this data in in a real time dashboard with the SAN performance data.
Vizioncore vFoglight is one of the market leading solutions for capturing, storing, alerting on, and reporting on the VMware vCenter data with a very attractive user interface. As such vFoglight is considered by many shops to be their foundation monitoring solution for the VMware environment. Vizioncore is also aÂ subsidiaryÂ of Quest, which owns the full Foglight assets (a full APM solution), and the recently acquired Surgient private cloud management platform. The combination of these assets position Quest to enter the Service Assurance space, once these assets are integrated into a cohesive solution.
vKernel started and leads that category of capacity management solutions for VMware. The product suite now includes freeÂ versions, optimization, capacity analysis, and chargeback. vKernel as also recently announced that it has added support for Microsoft Hyper-V to its chargeback and capacity analysis solutions. Managing performance and capacity across virtualization platforms will be important in 2011, and vKernel has staked out an early leadership position in this important new area.
VMTurbo is one of the two vendors listed here (Platform Computing is the other) that has an offering in the Performance Assurance space today. VMTurbo takes an extremely unique approach which involves allocating a budget of virtual dollars to workloads, and then pricing scare resources in those same terms. An economic market clearing model is used to determine who has enough money to pay for the scarce resources and ensure that they are allocated to their highest and best use. VMTurbo’s solution is today focused upon balancing out workloads in a virtual environment. As VMTurbo adds the ability to consider applications response time as the criteria for application service level, VMTurbo will become a very interesting solution.
VMware vSphere is the market leading and most technically rich and complete virtualization platform. However being the market leader in the platform, does not mean that VMware is or will automatically be the leader in any aspect of virtualization management. Today VMware offers a series of virtualization management products, most of which are the result ofÂ acquisitions, none of which are integrated with each other into a coherent management suite, and only some of which are even integrated into vSphere. On the configuration management front VMware offers vCenter Configuration Manager which VMware acquired from EMC/Ionix and which EMC had acquired when EMC bought Configuresoft. On the resource and availability management front VMware offers SpringSource Hyperic which comes from SpringSource buying Hyperic, and VMware buying SpringSource. Both of these products support both virtual and physical envionments. Hyperic has recently been integrated into vCenter, but this has not occurred with Configuration Manager yet. VMware also offers AppSpeed on the APM front which resulted from the acquisition of B-hive. AppSpeed is well integrated into vCenter, and provides APM functionality for virtualized (by not physical) web/J2EE/.net/database application.
Xangati comes at the Infrastructure Performance Management market from the perspective of the LAN and the WAN. Xangati uses deep and continuous instrumentation of the network to determine what is talking to what, and what the response times are for specific VM’s, hosts, and storage devices. Xangati has gotten some impressive early traction in VDI environments as customers have discovered that the end user experience problems that often derail VDI pilots are observable through the network and sometimes also caused by network issues – making Xangati a perfect fit for these situations.
Zenoss is an commercial open source vendor focused upon the resource and availability monitoring space. The core Zenoss product is a free open source download with a broad and deep community that continues to expand the scope of the product. The paid for version of the product includes the support for the VMware vCenter data, Cisco UCS, dynamic mapping of the environment and many other worthwhile features. Zenoss has also been proven to scale up to 25,000 devices and has been used by customers to replace legacy systems management frameworks.
How to Pick a Solution
How to pick a solution from the numerous choices above is a complicated topic best left to another article. However some basic guidelines are:
- Decide what problem you are trying to solve and for whom you are trying to solve it. For example if all you want to do is make sure that you do not run out of host CPU and memory capacity take a look at an inexpensive resource and availability monitoring solution that includes capacity management features. If you are trying to virtualize business critical applications and want to assure the performance of your infrastructure take a look at the Infrastructure Performance Management Solutions. If you want to help your applications teams understand the performance of their applications in a virtualized environment help them take a look at the APM solutions.
- Do not confuse VMware’s position as the vendor with the leading platform with any leadership on the part of VMware in the performance and capacity management business. VMware’s offerings are at this point undifferentiated (they have no features that other vendors do not have), disjointed (not integrated with each other with differing product architectures and user interfaces), and inconsistently integrated with vSphere itself. There is no doubt that VMware intends to be a major vendor in this area, but VMware is just not there yet.
- Balance a tactical (solve today’s problems with today’s budget) and a strategic (work toward the end goal of a self managing dynamic data center that assures the performance of business critical applications) in concert. This means try to purchase products that solve a real problem today, but that also will fit into your next generation strategic systems and applications management architecture. Try to avoid buying products that are limited to one or more vertical silos in your environment. For example avoid something that only works with one storage vendors array, one network vendor’s routers, for for that matter (from a roadmap standpoint) only one virtualization platform. Look for products that address an entire horizontal layer of requirements that that can be integrated with solutions in adjacent layers.