Virtualization Management

Ensuring Virtualization Performs

Effective virtualization management software ensures your virtual infrastructure performs as requires. To do this it provides visibility across your virtual and physical silos of servers, networks, applications and infrastructure. Because virtualization is dynamic, it changes the status quo relationship between servers, networks and applications. No longer is physical infrastructure provisioned for one application, or one desktop user, or one file system. With virtualization, applications can move between servers and end user desktops can be split into multiple components that reside most anywhere. This dynamic flexibility is the key way virtualization delivers its benefits – and management tools must be able to handle it.

“A completely different breed of management solutions is essential to manage the dynamic environment of today’s growing virtualized infrastructures. ”
- Vanessa Alvarez, Industry Analyst at Frost & Sullivan

An effective virtualization management solution must:

  • Work for servers, applications and VDI
  • Provide correlated visibility across silos
  • Show dependencies and interactions – both virtual and physical
  • Continuously analyze performance – second by second
  • Assist in designing, troubleshooting and optimizing virtualization

Server, Application and VDI (Virtual Desktop Infrastructure)

An effective virtualization management solution of course works across a majority of virtualization applications and supports a wide range of technologies. Let’s look at some of the specific things an effective solution needs to help with:

Server Virtualization

  • Auto discovering and identifying virtual machines, hosts, applications, end-users, switches, routers
  • Providing visibility into interactions between a VM or host and everything it interacts with in real-time
  • Real-time alerting and DVR recording activity to troubleshoot problems with full context into what was happening
  • Cross-silo visibility across virtual and physical infrastructure
  • Application aware – what applications are running and what do they need

Virtualized Applications

  • Discover all application dependencies across both physical and virtual infrastructure, ALL necessary components for an application to perform, including, but not limited to, presentation servers, application servers (aka business logic servers), databases/file systems, and heavily allied components including DNS and Active Directory services
  • Identify servers with the greatest dependencies to determine which specific application components should be running on the same host. This dependency information is the #1 information requested from VI administrators when creating cluster designs. Leveraging dependency information in designing VM placement results in better performance of the application on the VI due to better communication performance between the dependent servers.
  • Identifying bandwidth usage/requirements for applications including bandwidth reductions for VDI users of applications over WANs
  • Identifying and tracking end user usage of applications

Virtual Desktop Infrastructure

  • Evaluate and select the right VDI protocol for your environment
  • Continuously watch network latency as latency > 200 ms can kill VDI end user experience
  • Identify peak VDI workloads which can kill end user experience
  • Green-light applications for low capacity network connections and/or WANs
  • Certify remote sites viability of VDI
  • Tune the VDI protocol with real-time visibility into the affect of configuration changes
  • Analyze WAN bandwidth usage by VDI users, applications, web traffic and resulting positive or negative impacts

Provide Correlated Visibility Across Silos

Cross-silo visibility is critical as virtualization puts server, network and application performance responsibilities into the hands of the virtualization infrastructure administrator (VI admin). Server responsibilities as the VI admin creates new servers in just minutes as virtual machines on physical server hosts. Network responsibilities as network traffic between VMs may never leave the virtual environment – meaning the network team with their tools may never see or be able to resolve issues including confirming proper configuration of the virtual NIC cards associated with each VM. Application responsibilities as the way virtualization problems will arise is through applications that suddenly don’t perform as expected -- with virtualization getting blamed. Without a complete view, correlated by time, troubleshooting performance problems is a daunting and time-consuming endeavor.

Show Dependencies and Interactions – Both Virtual and Phsyical

High availability of virtual machines and hosts is not enough for the VI admin to ensure adequate performance. VI admins have to associate infrastructure to the applications it supports. Standard server metrics (CPU, Memory, Disk) don’t provide this information. To effectively manage virtualization VI admins need to know what VMs are dependent on. For example, is a VM part of an application with twenty or thirty VMs spread across several hosts and clusters? Does this application have backup and recovery jobs? Are they on dedicated VLANS? Where are the users of this application located – and what are typical usage patterns? Without visibility into dependencies and interactions – a VI admin is nearly powerless to resolve problems blamed on the virtual infrastructure whether or not the cause is virtualization related.

Continuously Analyze Performance – Second By Second

Virtual infrastructures are dynamic. VMs can move between hosts in under a second. Traditional polling solutions that look at performance every 15 minutes or even every 5 just don’t add value in virtual environments. Effective management solutions analyze performance activity on a second by second basis. By analyzing performance continuously - dynamic problems are quickly identified and can be quickly resolved. For example, if a VM has a surge in activity – a large file transfer, database query or backup job, this surge in activity can impact all the other VMs on that host – or sharing the same virtual network. These activity surges may last only seconds or minutes – but solutions that look at activity every 5 or 15 minutes will never see these and not help resolve them. Without continuous analysis – virtualization can be blamed for many things it has nothing to do with – needlessly slowing deployments and reducing ROI.

Assist in Designing, Troubleshooting and Optimizing Virtualization

Shouldn’t an effective management tool help you with all stages of virtualization – from designing a deployment to troubleshooting it and then optimizing it? We think it should. DealFlow Networks works with Xangati because it helps with all stages – from server to application to desktop virtualization:

  • Designing -With detailed virtual and physical communication profiles covering CPU, memory, storage latency, application usage, and more, Xangati enables you to design superior virtual deployments from day one. Xangati discovers and presents the comprehensive set of dependencies for any application or desktop, from active directory servers and DNS to database and application servers to presentation servers and storage. Dependency information is the number one request for cluster design; with communication profiles, you can ensure your top talkers are appropriately located for the performance you desire. And, with visibility into communications across both virtual and physical infrastructure, you can identify weak points or bottlenecks that can rob performance. WAN congestion? vSwitch or NIC problems? Poor storage performance? With Xangati you see it and are able to take action. Is usage of a particular asset or application rising? Use Xangati to track usage and design your deployment to handle it without guessing and without relying on averages that deliver nothing except trouble when designing systems to perform reliably.
  • Troubleshooting -Live, continuous visibility into detailed communications and resource usage enables problems to be identified in minutes that would otherwise take hours or days (if they could be identified at all). By incorporating drill-down navigation into its dynamic UI of live communication, Xangati enables you to walk activity across your virtual and physical infrastructure just by clicking your mouse. Each click brings up anything and everything communicating with the identity you are looking at—be it a desktop, application, VM, port, protocol, host, or NAS. Within seconds you will know if a VLAN is misconfigured or if iSCSI storage traffic is competing with something it should not. Xangati calls this VM-to-Anything” visibility. If users complain about performance, Xangati will immediately show what they are doing and identify if the problem is with their activity (are HD videos crowding out their email or applications?), the infrastructure or an application. Showing actual activity is critical because it includes peak activity—something missed by polling-based solutions that can only show averages or a few moments in time over the course of an hour. And, intelligent profiling enables Xangati to send alerts and/or trigger automatic recordings based on out-of-profile communications and/or resource activity. When trouble starts you have a DVR recording of exactly what was happening at the time. No more problem re-creation; you have a video showing what was going on at the time of the problem. Manual recordings can also be made with the push of a button or scheduled in advance to catch “hiccups” that occur routinely, but without known cause.
    • Troubleshooting Diagram
  • Optimizing -Optimizing performance is easy when second by second analysis of your virtual and physical infrastructure is presented in the same UI. You no longer need to access multiple systems to see what is actually going on. With Xangati, information is pulled from vCenter, vSwitches, and even physical switches and routers for profiling so that DVR recordings are made automatically when activity is out of character. This enables performance to be optimized for a wide range of conditions. For example, DRS affinities can be set and updated based upon changing VM behavior, allowing business-critical applications and virtual desktops to perform optimally across the entire enterprise. Monitoring solutions cannot deliver the granularity needed to optimize performance—the reason high-performing customers add Xangati regardless what their monitoring tool portfolio looks like.

© 2011 Deal Flow Networks. All Rights Reserved.