Whitepaper: The Top 12 Requirements for a Complete Virtual Lab

Surgient just released a new interesting whitepaper:

To meet the demand for higher levels of service at lower cost, organizations are evaluating virtual lab automation solutions to extend the power of virtual infrastructure software and reduce the time-to-benefit of key application development processes. In this white paper, youll learn about the critical capabilities that must be present to have a complete virtual lab automation solution. Regardless of whether a virtual lab is managed by a packaged commercial application or by in-house developed technology, the lab must provide the top 12 capabilities discussed in this white paper. With this information, you can more easily evaluate and choose from among different virtual lab solutions.

Download it here.

Interview: virtualization.info interviews Massimiliano Daneri

Hello Massimiliano and thank you for joining us on virtualization.info.
You are worldwide famous for your Perl script VMBK, able to perform virtual machines live backup in a VMware ESX server infrastructure.
Now you just released the first beta of a new tool, VMCL, which seems even more exciting.

VI: Can you explain what VMCL does?
MD: VMCL is the first free high availability software for VMware environment, It’s based on famous Linux-ha cluster software.
It manages virtual machines failover when a physical host crashes
VMCL has a feature actually not available in any other commercial product: it decides which available physical host has to start faulty virtual machines depending on specific parameters (free memory, CPU usage) to be respected and considering a defined priority.

The whole thing is based on a quorum disk (a VMDK formatted ext3) created for every physical host, where every virtual machine configuration file (.vmx) is copied and synchronized.

When a physical host is faulty VMCL will decide which remained physical host has to mount faulted host quorum and to start which virtual machine.

VI: To achieve such results with kind of infrastructure we’ll need in terms of networking speed and storage capacity?
MD: You need very few resource: 1 network adapter for heartbeat if you don’t use console network adapter, 100 MByte for the quorum disk on a VMFS volume for each physical host.
Anyway VMCL is a porting of Linux-HA project so you could use a lot of feature of Linux-HA, like more than 1 network adapter for heartbeat, a serial port for node communication, and usage of other resource on cluster (like virtual IP).

VI: How much data is moved between synchronized hosts in typical scenarios?
MD: You have 2 synchronization process, one is the heartbeat, this is managed by Linux-HA (you can choose between a multicast/nicest/broadcast intranodes comunication) while the other one is the .vmx synchronization between the physical host and the quorum disk. This last operation occur every 2 minutes with a cron job, so we are talking about very few bytes.

VI: As you know other companies are offering this feature for ESX Server. The last one is Vizioncore expected to release esxReplicator soon. Why did you choose to distribute this tool for free when it’s evident there is a market for it?
MD:
I used to work with Veritas Cluster for VMware ESX Server on a complex virtualization infrastructure of 72 ESX Servers and I think I have a clear idea of what a cluster software have to do and what not. I’ve encountered so many problem with Veritas that I had to rewrite all VMware scripts.
VirtualCenter integration and performances were a disaster but now, with a custom agent, it works well.

The major problem of this kind of implementation is treating a virtual machine like a proccess, which can freeze cluster resources.
So I decided to create VMCL with a different design and to distribute it for free cause I’m a consultant and I sell know-how, not products.

VI: Do you plan to offer VMCL also for the upcoming VMware Server running on Linux and Windows hosts?
MD: I’ll release a version also for VMWwre Server on Linux hosts. For Windows it’s more complicated since I would need to port the Linux-HA project and it’s very time consuming. Maybe I could accommodate it for Microsoft Clustering Services.
I also plan to release a VMCL version for upcoming ESX Server 3.0.

VI: Can you provide an eta for the final release of VMCL?
MD: In the upcoming days I’ll release the beta 2 with support for more than 2 physical hosts, featuring a new automatic setup.
I hope to finish VMCL 1.0 for middle April and show it at VMware Technical Solution Exchange (TSX) in Paris.

VI: Can we expect in near future an integration of VMBK and VMCL?
MD: I’m thinking to use Minime project library to create a graphical interface for both VMBK and VMCL.

As a scoop for virtualization.info I tell you I’m implementing a smart copy for VMBK where only modified .vmdk blocks will be transferred.

Akimbi joins VMware Community Source program

Quoting from the Akimbi official announcement:

Akimbi Systems, the global leader in Virtual Lab Automation Solutions, today announced it has joined the VMware Community Source program, an innovative collaborative development effort that provides open technology access to the source code and APIs for VMware ESX Server. The program combines the best elements of open source and enterprise development allowing partners to boost the level of functionality and slash the time to market of virtualization solutions for their customers. Akimbi will use access to VMware source code to optimize Akimbi Slingshot’s capabilities, performance and ease-of-use for deployment alongside VMware virtual infrastructure.

Akimbi will release an upgrade to Akimbi Slingshot incorporating features and performance gains enabled by VMware Community Source membership in the second half of 2006.

Xen 3.0.2 to be released this week

Anthony Liguori is confident we’ll have Xen 3.0.2 sometimes this week:

I dropped my latest (and hopefully final) version of the XML-RPC enablement for Xend this afternoon. The XML-RPC support is one of the last features to be merged before the 3.0.2 release so I expect we’ll see 3.0.2 go out sometime this week.

I’m quite happy that we’re actually keeping to our 4-8 week release cycle. Unfortunately, we’re probably going to slip a tiny regression into the 3.0.2 release. Last week, I noticed that our block-attach code is failing…

Read the whole post at source.

Release: PlateSpin PowerRecon 2.0 released!

Quoting from the PlateSpin official announcement:

PlateSpin Ltd. today announced the general availability of its next generation data center analysis and optimization solution.

PlateSpin PowerRecon 2.0 remotely collects a wealth of relevant hardware inventory data, resource utilization and workload statistics that create the foundation for one-time consolidation or continuous server optimization projects. It allows users to remotely identify what resources exist in the data center and presents a complete hardware, software and services inventory with absolutely no manual effort or disruptive agent technology required. When used in conjunction with PlateSpin PowerConvert, data center users and consultants can identify consolidation candidates, and perform fully automated migrations from a single product suite.

PowerRecon 2.0 features:

  • identify under-utilized and over-utilized servers quickly and easily
  • generate time-based graphs which include individual or combined server workload, allowing users to consolidate servers that have staggered workload characteristics
  • visualize host and virtual machine hierarchies for Microsoft and VMware virtual platforms
  • completely agentless architecture eliminates the need for software to be installed on any data center servers
  • measure and analyze over 1,000 servers with minimal impact on network traffic
  • completely new graphical interface provides users with highly effective visualization of workloads and resources
  • customizable resource and workload reports for speedy capacity planning analysis


PlateSpin PowerRecon 2.0 is generally available today through our global partner network and is packaged on a per-use and on a per-server basis to offer our customers and service provider partners licenses that best fit their needs. Per-use pricing starts from $1 per day for each individual server. PlateSpin is also offering perpetual PowerRecon licenses and specially priced bundles when purchased together with PlateSpin PowerConvert.

Tech: Network Load Balancing and Microsoft Virtual Server 2005

Ben Armstrong explains once and forever issues with Windows Network Load Balancing (NLB) and Microsoft Virtual Server 2005:

…Virtual Server / Virtual PC will not utilize any NLB capabilities on the host operating system. This is because Virtual Server / Virtual PC perform their network operations at a lower layer on the host network stack than NLB. Note – Virtual Server and Virtual PC will function correctly when NLB is present on the host – they will just not benefit from it.

Virtual Server / Virtual PC have some network optimizations that actually cause problems when running NLB systems inside of virtual machines. These optimizations are simply that if we see a network packet that is going from the host to a virtual machine, or from a virtual machine to another virtual machine, we will send the packet directly to that virtual machine – without sending a copy of it on to the physical network connection. Similarly if we see a network packet coming into the physical computer that is destined for the host operating system – we will not send it on to any of the virtual machines…

Read the whole article at source.

Sun to launch public Grid

Quoting from BetaNews:

Sun will finally launch its much-hyped yet much-delayed public version of its Grid product later this week, its president said. Users in the United States would be able to purchase processing cycles at the rate of $1 USD per hour each on their choice of AMD Opteron or Sun UltraSPARC processors.

The company would initially be making 5,000 CPUs available for use, with plans to increase that number as necessary. Users would be able to access the grid from the Internet at Network.com. Payment for use would be accepted through PayPal…

Read the whole article at source.

Read more from the Sun President, Jonathan Schwartz, blog.

Novell to integrate Xen in SuSE Linux Enterprise 11 in 2008

Quoting from CRN:

Novell will include its “Beagle” search engine and virtualized server and storage features in SUSE Linux Enterprise 11, one Novell executive said.

As it officially launched SUSE Linux Enterprise 10 at Brainshare 2006 Monday, Novell vice president Holger Dyroff gave an overview of features planned for SLES 11, expected to ship in mid-2008.

For instance, SLES 11 will offer virtualization for servers and storage, enhanced security via support for AppArmor’s profile management and CIM, a compiler and run-time, and identity management and provisoning, Dyroff added…

Read the whole article at source.

For that time Microsoft could be (quite) ready to resease its Windows Hypervisor (codename Viridian) as add-on for codename Longhorn Server. So Novel should use these 2 years in the best way to offer a consistent Xen management infrastructure.

Bull announces Itanium 2 supercomputer and SWsoft Virtuozzo support

Quoting from the Bull official announcement:

Bull today announces the NovaScale® 5005 series, representing the second generation of large NovaScale servers.

“The NovaScale 5005 servers, announced three years after the first generation, are a proof of Bull’s innovation capabilities. This announcement is in line with Bull’s positioning as an Architect of an Open World. Running present and future generations of Intel® Itanium® 2 processors, they are naturally at the heart of the most complex IT infrastructures, both in the Linux and Windows open environments”

The combination of physical partitioning and virtualisation software makes the NovaScale® servers uniquely flexible to use and enables users to fully optimise the resources installed:

  • The very thin partitioning of the NovaScale 5005® servers – a minimum of 4 sockets per partition, compared to 8 with the previous series – makes it possible to partition a large server into two to eight servers within a few minutes. Each server being able to run a different application environment. On the other hand, multiple servers can be united into a single server to face an increased need in power for a given environment.
  • Bull also announces the availability of the SWsoft’s Virtuozzo solution for the Linux environments on the new NovaScale servers. The Virtuozzo virtualisation solution operates at the Linux operating system level to generate multiple private virtual servers, on a single physical server or on a physical partition…

Thanks to Virtuozzo Blog for the news.