Virtualization

You are currently browsing the archive for the Virtualization category.

In a previous post, I mentioned the announcement of the Sun’s Ops Center product targeted to the management of virtual environments. In this post, I said that Ops Center was a re-branded N1 System Manager, while in fact, it seems that this is a merge of the Sun Connection and N1 System Manager in one tool :

A highly scalable datacenter automation tool merging discover, update, provisioning, monitoring, and reporting technologies from Sun Connection and N1SM into one tool.

However, by looking at the Oracle World demo, it seems that the UI is radically different from the N1 System Manager (gone the embedded CLI ?).

Also, by looking at the supported platform, it seems that Windows platform is not supported anymore :

Sun N1 System Manager :

From a centralized management console, customers can provision Solaris, Linux, and Windows with a simple drag-and-drop, and monitors the health of systems in an efficient manner.

Sun Ops Center :

The comprehensive, highly scalable Linux and Solaris life cycle management tool.

The good news is that Sun Ops Center will be delivered as open source too :

Building on Sun’s commitment to open standards and customer choice, Sun will continue to innovate the Sun xVM platform and collaborate with open source communities. The first of Sun’s contributions will be the Common Agent Container (CAC) code to the OpenxVM.org community under GPLv3. The CAC is the heart of the management infrastructure for many of Sun’s products, including the Sun xVM Ops Center. In addition, Sun plans to make the entire code base used by Sun xVM Ops Center available to the OpenxVM.org community in the first quarter of 2008.

It’s not clear however if this means the end of life for the N1 System Manager, since right now, the Ops Center does not provide a complete replacement.

 

Sun announced new virtualization products last week. It seems to be a version of Xen hypervisor running on solaris . So now, Sun has three Solaris virtualization technologies: XVM, Solaris Containers, and LDOM. So, what it means is that a Sparc Server can be running a solaris container in a solaris XVM in an LDOM. This is 3 layers of virtualizations.

Also, in this announce, Sun presents Ops Center. This is a re-branded N1 System Manager. It seems that the focus will be on the management of the hardware, up to the operating system, combined with Sun Connection.

Mark Hamilton told The Register:

Sun has also announced a new management software package, Ops Center, that will work as a command and control console for physical and virtual gear- that’s to say the hypervisor and Solaris Containers. Sun said the software also includes discovery and inventory, application provisioning, software lifecycles automation, hardware and software monitoring and compliance reporting. Sun brazenly says “it does everything except unpack boxes and rack and cable systems.”

Ops Center is going to be released in December. It will be interesting to see how this can be leveraged by Configuration Automation tools like what BMC announced recently.

 

According to this article, Parallels was acquired 3 years ago by SWsoft:

The tectonics of virtualization are shifting. It turns out that Parallels is not such a little company after all. About three years ago it was quietly purchased by an enterprise-focused virtualization company called SWsoft, a fact that has never been publicly disclosed until now.

SWsoft is providing Plesk, HSPcomplete, PEM and Virtuozzo, which seems to be the only virtualization related technology. It will be interesting to see if SWsoft applies its expertise in Self Service, Accounting and Billing to the VM Management space.

Today, VirtualBox announced the release of their product in the open source, using the GPL:

Jan 15, 2007. InnoTek today released VirtualBox Open Source Edition (OSE), marking an important milestone in the development of PC virtualization software. VirtualBox OSE is the first professional PC virtualization solution released as open source under the GNU General Public License (GPL). With VirtualBox, customers get the most versatile virtualization product on the market, both for enterprise and individual use. VirtualBox’ open source license allows everyone to contribute to the development of the product and customize it to suit individual needs. Backed by Europe’s largest team of virtualization experts, VirtualBox continues to be developed and supported.

Going through the screenshots, you can see that they have an alpha version of VirtualBox for OSX :

vbox_osx_alpha

 

Today, the VMWare Fusion Beta program opened :

The new VMware desktop product for the Mac, codenamed Fusion, allows Intel-based Macs to run x86 operating systems, such as Windows, Linux, NetWare and Solaris, in virtual machines at the same time as Mac OS X. It is built on VMware’s rock-solid and advanced desktop virtualization platform that is used by over four million users today.

I’ve tried to start a windows 2003 VM Image that was created on a PC, without any problem.

fusion

I have been trying for the past few days to run Solaris 10/11 in parallels on a new macbook. The main issue I had was the JDS/gnome resolution adaptation with the 1280×800 resolution of the macbook.

Here are my steps :

  • I installed the new parallels beta (build 3036)
  • downloaded the free Solaris 10 update 3 from Sun’s web site. Download the DVD since it can be directly mounted in parallels.
  • create a new VM with the Solaris 10 type, but, before to finish, un-select the option to start the Solaris installation and edit the VM configuration to add a custom screen resolution of 1280×800.
  • start the VM to launch the install
  • log in an create a /etc/X11/xorg.conf with something like the one attached here (xorg.conf). Basically you need to add the Modeline lines (generated with /usr/X11/bin/gtf) and put 1280×800 in the appropriate Display subsections.

It should work. However, I’m still experiencing some duplicate keystrokes and garbled screen after VM resume (I have solved the later issue by switching multiple times between full screen and os window).

If anyone has any solution for these two problems, let me know.

Today, Scalent announced Scalent Virtual Operating Environment(TM) (V/OE) version 2.0.

With Scalent V/OE, infrastructure repurposing becomes nearly instantaneous. Changing system function and topology doesn’t require touching physical cables or machines. Instead of spending hours or days reloading software, changing configurations and even moving machines and cables, IT teams can rack once, cable once, then reconfigure repeatedly, in minutes, effortlessly. Failover is automatic, and data center functionality matches the data center schematic — what you see is what you get.

One issue with that approach is that datacenters are heavily silo’ed : the network is administered by a group different from the group managing the storage, or the servers. In the worst case scenario, these functions are even outsourced, either traditionally or leveraging technologies like Amazon S3. Is that something that Scalent is addressing ?

More and more, managing servers, and services, will become a business integration problem, using concepts usually reserved to e-commerce, implementing change management best practices, and leveraging business process automation tools.

In a previous post I tried to capture the functions required to manage datacenters and services across their lifecycle. In a subsequent one, I tried to map the VMWare virtual infrastructure 3 functions. Obviously, the granularity was not the right one. Below is another attempt to capture what the holistic management offering should be :

200606261336

 

Hardware Management

Not specific to the virtualization domain, but critical to have an end to end solution. Usually, this part is implemented with solutions like IBM Director, N1 System Manager, or HP Insight Manager. However, one of the many issues when done in a virtualization setup is the control of virtual machines in a way similar to physical machines (with a virtual IPMI implementation for example).

  • HW Discovery: This is discovery of bare metal systems using IPMI or other vendor specific baseboard interfaces. The machine does not need to have an OS, or Domain 0 to be discovered.
  • FW Update: Update of the BIOS or other hardware specific software. Usually requires some tight integration with hardware management interfaces.
  • HW Monitoring : Monitoring of environmental information like fan speed, CPU temperature …
  • Hardware Abstraction Layer : A kind of driver layer allowing to decouple the hardware specific functions from the various protocols or vendor interfaces. Critical to be able to, for example, configure the machine to boot by using PXE.

Hypervisor

the central piece of the virtualization offering, this could be Xen or VMWare ESX for example. This layer is now a commodity and offered for free as part of the OS.

Virtual Server Management

Functions usually implemented in the Domaine 0 (aka Host OS, or Service Partition).

  • Virtual Device Management : Management of devices exported to Virtual Machines. This includes network, disks, PCI devices, consoles, … this function is usually highly tied to the hypervisor and share its implementation with the virtual server management. Can include the fault management of the various devices.
  • VM Management: Life cycle management of the Virtual Machines instanciated on a given host. Provides essentially methods to boot VMs, either from local disk, CDROM or network (e.g. through PXE).
  • Volume Management: Provide a logical volume manager, as well as various storage related functions.
  • Resource Management: Allocations of physical resources to many VMs hosted on a given server. Includes the scheduler configuration.
  • Monitoring: Monitoring of resources used by various VMs. Could be an aggregation point for information collected in the various guest OS.
  • Migration: Live migration of VMs from a server to a given target server. Could be more or less stateful or real time dependending on the virtualization layer capabilities.

OS Management

Mainly targeting the management of the guest OSes, and therefore not specific to the virtualization domain, but critical to have a end to end solution. Could be implemented by solutions like IBM Director, N1 System Manager, or HP Insight Manager but this would lead into additional complexities introduced by the management of virtual machines.

  • OS Provisioning: Infrastructure required to implement NAS boot or SAN Boot, as well as building OS profiles
  • OS Monitoring: Monitors OS parameters such as CPU utilization, directly from the guest OSes.
  • OS Patching: Update of OSes with patches. Additional complexity when required to update non deployed VMs.

Datacenter Management

  • Datacenter Resource Management : Grouping of server, storage and network resources and allocation to various services. Basis of multi-tenant aware solutions, or hierarchical resource management capabilities. Can implement some level of workload management.
  • Failover : migration of virtual machines based on availability related policies.
  • Configuration management : repository of various information about virtual machine parameters, or OS configurations. Could be also a repository for OS profiles, OS versions, …
  • Policy Management : More generic policies driving the operations like migration, or provisioning. Can be linked to performance goals, cost/utilization goals, schedules (calendar events)
  • Automation : Definition of workflow on top of basic operations allowing the implementation of complex operations.

Next step is to try to map the various vendors to this breakdown.

Last week, VMware announced their acquisition of Akimbi and their Slingshot product:

VMware adds Virtual Lab Automation, Configuration Management and Self-service Provisioning to its Portfolio

PALO ALTO, Calif., June 16, 2006 — VMware, the global leader in virtual infrastructure software for industry-standard systems, today announced the addition of virtual lab automation, configuration management and self-service provisioning to its software lifecycle management solutions that span development, quality assurance, pre-production and production environments.

If you remember my last post about Virtual Infrastructure 3, you can see that this acquisition would complement nicely Virtual Infrastructure as there was some gaps in the provisioning area.

On June 5th, VMWare announced a suite of products under the label Virtual Infrastructure 3.

The components of the suites are the following :

  • VMWare ESX Server : ESX is the virtualization layer installing on top of the Bare Metal. It seems that ESX is able now to run VMs created by Microsoft Virtual Server, or Virtual PC, or also Symantec LiveState Recovery images.
  • VMWare VMFS : VMFS, the Virtual Machine File System is an extension of the ESX Server allowing multiple VMs to read and write from the same file storage location. This file system allow adaptive block sizing and automated LUN discovery and mapping.
  • VMWare Virtual SMP : The virtual Symmetrical Multi Processor is an extension of ESX Server allowing a single VM to use up to 4 physical processors (within the same physical machine). The scheduler within Virtual SMP allow over commitment of the processors as well as reuse of idle processors between VMs.
  • VirtualCenter: This is the centralized management server enabling features like VMotion, VMWare HA, VMWare DRS, VMWare clustering, resource pools, … which are really targeting enterprise class deployments.
  • VMWare Virtual HA: Virtual High Availability is a simple failover solution using a heartbeat to detect hardware failures. Virtual HA will restart an identical VM on another machine. However, there is no mention of checkpointing of the VMs to ensure minimal state loss. This feature is integrated with VMWare DRS to enable intelligent VM placement.
  • VMWare DRS: The Distributed Resource Scheduler allow the definition of advanced resource allocation policies for distributed applications. DRS enforce these policies at initial placement time by selecting the best server for a given workload, or continuously, to optimize the resources allocated by eventually moving the VMs through VMotion.
  • VMWare VMotion: allow the migration of VMs across all types of physical servers supported by ESX, including across Fiber Channel SAN, NAS, or iSCSI SAN storage devices.
  • VMWare Consolidated Backup : It allows the backup of virtual machines from a central location. This is a set of scripts allowing that enable LAN-free backup from a centralized Microsoft Windows 2003 proxy server.

VMWare has segmented the market in 3 categories targeted by a specific offerings :

  • Small Business or Branch Office with VMware Infrastructure Starter. It includes ESX starter and VirtualCenter management agent. It retails for $1000 for each couple of processors (physical processors, not cores).
  • Enterprise class infrastructure with VMware Infrastructure Standard. It adds clustered VMFS and virtual SMP. It retails for $3750 for every couple of processors.
  • Enterprise class Dynamic Datacenters with VMware Infrastructure Enterprise. It adds VMotion, DRS, HA and consolidated backup. The reatil price for this version is $5750 per couple of processors.

The following products are still available separately : VirtualCenter, VMotion, VMWare HA, VMWare DRS and Consolidated Backup.

If we try to map the resulting offering on the functional breakdown I previously described here, we get the following result :

200606141312

The rational for the coverage is :

  • Resource provisioning is partial since VMWare doesn’t seem to provide a solution to do the initial provisioning of VMs : You can use PXE to boot from the network, but you will have to manage the OS installation either through Microsoft RIS, or Kickstart.
  • Resource Discovery is also partial since VMWare doesn’t seem to provide a solution to discover bare metal servers by directly talking to the baseboard. Installing ESX Server from the CDROM or SAN would require a third party management solution.
  • Performance management is partial since the whole OS instance is monitored, and any metric available at the middleware or service layer would not be visible to DRS, VMotion of other VMWare products.
  • Availability management is partial also as only availability from the OS down to the hardware would be considered, and failure of the application or service would not generate the restart or migration of the VM (not that this would actually be able to fix the problem)
  • Workload management is partial since the VMWare solution is not taking into account the scheduling of transactions or workload items to make decision, but seems to only observe OS and hardware resources consumption. It’s not clear for example, how VMWare DRS would integrate with a load balancer or other middleware managed schedulers (like Oracle RAC).

Updated on 06/14/2006 to remove the Orchestration support since there is no modeling of worklows in VMWare.