Yesterday, Matt Liebowitz posted an article titled Fix Partition Alignment with VMware Converter 5.0 (Beta). He covers how the new Beta of the VMware vCenter Converter Standalone 5.0 can handle optimizing alignment inside a VMware VM.
This is a feature that VMware Admins have wanted for a long time.
What is alignment? Why is it important?
Duncan Epping posted an article that explains why guest OS alignment is important. You can find it here: Aligning your VMs virtual hard disks
My question is… Is correcting the alignment good or bad? Keep in mind the process, time, & costs associated with doing it. I might get some flak on this, but I think it is a valid question.
Referencing Duncan’s article, and the VMware paper Recommendations for Aligning VMFS Partitions that Duncan references, guest alignment can:
Track alignment for both physical machines and VMware VMFS partitions yields I/O performance improvements such as reduced latency and increased throughput. Creating VMware VMFS partitions using the VI Client results in a 64KB-aligned partition table and provides the foundation for a best practices storage layout.Track alignment for both physical machines and VMware VMFS partitions yields I/O performance improvements such as reduced latency and increased throughput. Creating VMware VMFS partitions using the VI Client results in a 64KB-aligned partition table and provides the foundation for a best practices storage layout.
Also, remember that this affects Windows guests that are running any Windows OS prior to Windows Server 2008 or Windows Vista. That being said, Windows Vista/7 and Windows Server 2008/2008 R2 are not affected. Linux distributions are also affected.
Now back to my question. Is it good, bad, or indifferent to align guest OS installations?
- Older systems that have been around for a long time (Windows 2000/2003/XP) can have reduced latency and increased throughput.
- Older systems can stay around longer, without having to worry about latency/throughput, resulting from misalignment.
- Current, misaligned systems, can be corrected to perform appropriately.
- Many systems (either P2V or V2V) can be scheduled to be converted.
- Systems that are misaligned will have some downtime/data migration issues to deal with.
- Older systems that have been around “forever” may stay longer then they really should.
- In a very large environment, the conversion process could take a very long time.
- How much performance increase will be observed?
Pay close attention to the VMware paper referenced earlier. And with the “it depends” disclaimer, your mileage may vary.
- How how many hours ($$$) will administrators require to complete conversions?
Could the hours an administrator is working on this, be better spent on a more critical business need?
- Will a conversion project impede other, more critical project, for a possible/expected performance increase?
Again, your mileage may vary.
I ask these questions, because I worked in an environment with around 600 VMs, and the expectation to scale to 1,200 VMs. When presented with an alignment tool, we made the decision NOT to align our older VMs, running Windows 2Kx, given that we knew they would be replaced by Windows 2008/2008 R2 systems in the foreseeable future. We did take care of alignment whenever any NEW (2Kx) systems were installed or P2Ved, as a best practice. The approval to align a new/P2Ved Windows 2Kx system was often a case by case basis.
Personally, I think this is an awesome addition to VMware Converter. I can’t wait to use it.
Just one more question though. What about that one system, for whatever reason, that really needs to just go away? You know, the one that is difficult/expensive to manage? The one that was setup by an admin that is no longer with the company, and no one really knows how it works. With it misaligned, there was another argument to put it to pasture and upgrade/update. Alignment can’t be a factor in getting rid of it anymore.
Will the benefit outweigh the time and cost? Or will the time/cost outweigh the performance gains?