Gerben's Blog on Virtualization

New blog post: http://tinyurl.com/8aofvo - VMware Virtual Infrastructure 4 will be named vSphere

This the final note to let you know that this blog (Gerben's Blog On Virtualization) has moved to virtualarchitect.nl. Please alter your links to contain the new url.

The feed address has also changed: http://feedproxy.google.com/gerbensblogonvirtualization
Existing feedburner subscribers will automatically be redirected to the feedproxy from google.
Unfortunately subscribers to the live.com-feed have to change the feed subscription manually.

The new blog is a self hosted Wordpress blog, which gives me more flexibility for appearance and functionality. All the content of the live-space has been moved to the new url.
Happy reading in 2009!

Merry Christmas and a Happy New Year!

Gerben

Lefthand Networks - John Spiers

John Spiers co-founder of Lefthand Networks (LHN) elaborated about Virtual SAN's in the Virtual Data Center. He started to mention the acquisition of Lefthand Networks by HP and showing a HP branded presentation. LHN was founded in 1999 and it was then they started selling ISCSI-based storage, even before the protocol was formally standardized, according to John Spiers.

LHN addresses problems like the fact that servers nowadays are mapped to static resources and the demand of virtual performance for ports, zones and RAID groups is rising, etc. How does LHN address these problems.
LHN has five (5) main features:

1. Storage Clustering
2. Network Raid
3. Thin Provisioning
4. Snapshots
5. Remote Copy (Disaster Recovery)

I was sitting next to a former colleague/customer, and while listening I remembered we recently discussed the concept of High Available an physically separated  Sites (+50 km) and the ability to vMotion a VM from one site to another. Almost every requirement was met to make it work except for the storage.

And I mean really high available and not a DR solution. When the first DC fails the other DC has to take over using HA or the new Fault Tolerance feature.

The discussion focused on the storage side and questions like how to make the storage high available in a way that the storage is available instantly in the other datacenter without having to act on it.

First another colleague mentioned Storage vMotion. We concluded that using Storage vMotion creates a low available solution (takes time to copy data with SvMotion) and you will end up failovering by hand if you're lucky to be able to do that in case of failures. So for this scenario it's a no go.

Site Recovery Manager (with underlying storage replication) is not a solution either, using SRM you still have to get the LUN out of replication mode to use it (make it active). Besides this, SRM is not an High Available solution it is a Disaster Recovery Solution and it takes downtime as granted. Someone has to push the button to failover with SRM, so another manual factor.

During this session another option popped up. Lefthand Networks offers a so called Network Raid, with the Network Raid it is possible to use a single SAN across (physically separated) sites. RAID 0 mirroring might be a worth a shot. Downside of this can be the network latency, this must be less than 5 ms. Also hardware dependency may be something to keep in mind

VirtualGeek wrote an interesting article about Storage vMotion partly related to the discussion mention above: http://virtualgeek.typepad.com/virtual_geek/2008/12/real-world-experiences-using-storage-vmotion.html#more

Then the session and the discussion ended... 

Comments are welcome.

* This BLOG will soon move to: http://blog.virtualarchitect.nl/ (already online) please alter your links *

* Please alter your feed subscription to: http://feedproxy.google.com/gerbensblogonvirtualization *

* Thank you! *

A couple of months ago Veeam acquired nworks. Nworks is known for its agentless connectors for managing an VMware ESX (also ESXi !!) environment in SCOM and HP Operations Manager (Openview). Veeam just released version 4.0 of the management connectors.

Nworks is one of the few connectors which integrates with SCOM & HPSOM. Especially in larger organizations where SCOM or HPSOM is the main monitoring tool, integration with the mentioned monitoring tools is more or less mandatory when introducing a new system (in this case VMware).

Using plugin's like nWorks in one way makes life easy for administrators, they don't have to write their own management packs and or scripts (in case of MOM), in the other way you have a challenge filtering the right events, because it generates a lot of events. But as they say better too much than too little.

Of course there are more monitoring tools like Veeam's own Monitor product and of course Vizioncore's vFoglight (formerly vCharter Pro), but with one monitoring solution in place introducing a second tool for monitoring is a political challenge in most cases, unless there is a very good integration with the company 'standard' tool.

The new release 4.0 is a major release, which means that you will have to upgrade all components (for existing users :)).

What's new in this release:

• Hardware monitoring is now available through VMware’s Virtual Infrastructure SDK
• Changes to performance metrics (changed and new ones)
• Improved VirtualCenter 2.5 API support
• VMware tools version events
• VMware datastores are now displayed in a separate branch of the topology
• The collection interval for the collector is now adjustable (can be increased)
• suppression of events in case of unknown previous state, which is great in larger environments so your monitoring tool will not be 'spammed' with events
• A single collector install file is now used for both the Microsoft MP and HP SPI
• The list of ESX hosts managed by the VirtualCenter is updated every collection cycle. In previous versions, the list of hosts in the Configuration UI was only updated when a user clicked “Validate”; now, the list of hosts under a VC server connection is updated as soon as the collector detects a change. The change is actually in the collector service, but the effect is visible in the Configuration UI.
• and more...
  Microsoft SC Operations Manager:
  http://veeam.com/vmware-esx-monitoring-ms-operations.html
  HP Software Operations Manager:
  http://veeam.com/vmware-esx-monitoring-hp-operations.html

* This BLOG will soon move to: http://blog.virtualarchitect.nl/ (already online) please alter your links *

* Please alter your feed subscription to: http://feedproxy.google.com/gerbensblogonvirtualization *

* Thank you! *

VKernel just released Capacity Analyzer 3.0. In addition to showing bottlenecks in hosts, clusters, and resource pools, it now shows bottlenecks in VMs -- both inside and outside of VMs. Even small environments have hundreds of VMs, causing ongoing monitoring and finding problems in each individual VM to be very time consuming.

Capacity Analyzer 3.0 sorts through them to show the most problematic VMs first, so you can quickly see which VMs are having current performance problems. With predictive analytics, we also identify which VMs will have performance problems and when, so that you can take necessary proactive measures. Download and try Capacity Analyzer 3.0 here in your environment.

We have also made over 20 improvements and fixes in Capacity Analyzer 3.0. For example, the new UI is now tabbed-based making it easier and faster to find the information you need.

The 14-day free trial of the fully functional, production ready Capacity Analyzer 3.0 is now ready for download

* This BLOG will soon move to: http://blog.virtualarchitect.nl/ (already online) please alter your links *

* Please alter your feed subscription to: http://feedproxy.google.com/gerbensblogonvirtualization *

* Thank you! *