Virtuozity

Link: http://nss.cs.ubc.ca/remus/

Remus is an open source project which provides fault tolerance for virtual machines (VMs) running on the Xen hypervisor. The actual release 0.9 works with tip of the xen-unstable repository, supports paravirtualization and hardware VMs in various 32- and 64-bit configurations for Windows and Linux.

Remus provides transparent, comprehensive high availability to ordinary virtual machines running on the Xen virtual machine monitor. It does this by maintaining a completely up-to-date copy of a running VM on a backup server, which automatically activates if the primary server fails. Key features:

• The backup VM is an exact copy of the primary VM. When failure happens, it continues running on the backup host as if failure had never occurred.
• The backup is completely up-to-date. Even active TCP sessions are maintained without interruption.
• Protection is transparent. Existing guests can be protected without modifying them in any way.

I am looking forward to the day Remus will be announced stable and hopefully integrated into XenServer sooner than later!

Link: http://www.vmware.com/products/fault-tolerance/

Fault Tolerance (FT) is one of the new VMware products that will come out in 2009, and from my point of view it's one of the most exciting. So what does it do? To put it simple, FT provides Zero-Downtime VMs with a single click. And that is a real bargain compared to VMware HA.

So what's the story with FT? In the past I had some really unpleasant experiences with clustering software. The first problem is that most approaches are pretty proprietary, which makes most solutions difficult or unintuitive to manage. Mistakes are almost preprogrammed if you put them into production too quickly. The second disadvantage is the need of additional hardware resources. And the third, but most important drawback, is the famous "Split Brain Problem" all clusters should be able to deal with. Maybe you agree with me that most available cluster solutions on the market today are not a piece of cake to deal with, but require serious training and experience.

But now there is VMware FT and deals with it all? Somehow, yes, it does. It makes proprietary clustering software obsolete by simply cloning the whole VM on to a second one in real time. The VM is unaware of being "clustered", because the second box is an exact replica hold on standby. If one host fails, the secondary VM is automatically activated. VMware calls this technology "vLockstep".

So you will get rid of your proprietary cluster software and you can even make mission-critical applications high-available that were never designed to be clustered. Is this the end of all other clustering solutions? No, it's not, because VMware FT only takes care of the box, not of the application, just like VMware HA did. So if you reboot your primary VM, so does the secondary one, because they are completely synced.

Don't get me wrong, I still think this product is great news and delivers what HA promised, but there are serious limitations and it will definitely not make application clustering obsolete. And we still have to deal with "Split Brain", because someone has to decide which host is down, which one is alive and which VM is the active one. At the moment I guess the algorithm will be the same as in HA.

Link: http://sourceforge.net/projects/vip-svmotion/

There is a nice SourceForge Project for the neglected SVMotion feature. Since VMware does not provide any integration into the vCenter GUI, Schley Andrew Kutz developed a vCenter plugin on his own that delivers if you don't like the command line. Great stuff, check it out!

Link: http://www.hypervizor.com/2008/12/video-tutorial-vmware-srm-01-installation/

Dieses vierteilige Video Tutorial von Hany Michael bietet einen guten Einstieg in den aktuellen VMware vCenter Site Recovery Manager. Es deckt folgende Themen ab:

• SRM Installation
• SAN Setup using LeftHand VSA
• SAN Replication
• Configuring the SRA and completing the SRM setup

Link: http://www.vmware.com/files/pdf/solutions/08Q4_VM_Exchange_Server_2007_VI3_WP.pdf

VMware hat eine Machbarkeitsstudie zum Thema "Virtualisierung von Exchange Server 2007" veröffentlicht. Auf der Basis von HP Blade Servern und EMC Storage hat man dort eine ausfallsichere Umgebung mit einem Konsolidierungsverhältnis von 4:1 geschaffen (40 VMs auf 10 ESX Hosts). Die Studie beschreibt vor allem den grundsätzlichen Aufbau der Umgebung, die Geschäftsanforderungen und besonders ausführlich den kritischsten Faktor Storage.

Das Referenzprojekt von VMware soll Kunden motivieren ebenfalls Exchange zu virtualisieren. Nachdem viele Kollegen bereits schlechte Erfahrungen mit der Virtualisierung von MSSQL Servern gemacht haben, scheint der Exchange Server nicht unbedingt einer der besten Kandidaten für eine virtuelle Infrastruktur zu sein.

VMware beschreibt das Projekt als einen vollen Erfolg und führt folgende Gründe für eine virtuelle Exchange-Infrastruktur an:

• Höhere Verfügbarkeit ohne die Komplexität eines Microsoft Clusters
• Verbesserte Skalierbarkeit
• Optimale Ausnutzung der Infrastrukturkapazitäten
• Vereinfachtes Management (Testing, Provisioning, etc.)
• Senkung der IT-Kosten
• Hardwareunabhängigkeit