Looking for a reliable hypervisor?
Every time a vendor releases a new hypervisor, it’s always more reliable and stable than all the previous versions and certainly better than the competition. But was the previous version all that bad?
Of course with the addition of more and more advanced features it becomes more and more difficult to create a reliable and stable product simply because of the immense amount of code and the interdependency between all the different components. VMware has always had a dedicated hypervisor and with the release of ESXi they further reduced the code base of vSphere ESXi to minimize the attack surface. This also reduces the amount of code to patch which further improves reliability, stability and security.
Last week I visited a customer who, much to my surprise, still ran some VMware ESX 2.5 servers. According to the customer the servers ran fine and they had never had any problems.
When we opened the console, we found out that these ancient ESX 2.5 servers have had an amazing availability during the last 4 years.
Check this out!
This server has been running for 1.460 days and 16 hours (4 years). In that time it has delivered a stunning 99,928% uptime!
In those 4 years, it has only had 25 hours of downtime.
When we checked, the system had run for 7 months straight without any problems.
The longest uptime this system has delivered during the 4 year lifespan is 16 months/1.3 years!
Very very impressive for an ‘ancient’ hypervisor which has been released in November 2004!
But what was the reason for that downtime?
9.6 of the 25 hours of downtime was scheduled downtime caused system halts or reboot for system maintenance.
15.6 of the 25 hours of downtime was caused by unscheduled downtime.
But what caused the unscheduled downtime? A system crash, PSoD, etc.?
No, the unscheduled downtime was caused by an external cause, they suffered 4 power failures.
True, ESX 2.5 does not have the advanced features which the current hypervisor do, but it had the enterprise hypervisor basics like HA, vMotion, DRS. Features which some major players in the hypervisor space did not have a few years ago.
So the next time when you’re challenged on reliability, availability and proven-technology, look back at what VMware could offer you 9 years ago and how the product has evolved over the years. I dare you to post an ‘untempered’ screenshot of a Hyper-V version 1 or 2 with an equal amount of uptime.
Related Posts
6 Comments
Leave a Reply Cancel reply
You must be logged in to post a comment.
Meh… Couldn’t even get a year straight, try these up times http://vmguy.com/wordpress/index.php/archives/309
I was wondering what tool and method was used for determining the uptime? I’m still looking for a good tool.
This is the integrated ESX management which reports uptime and statistics.
I don’t care about hypervisor uptime, I care about my Virtual Machine and application uptime and performance.
Of course that’s true but you can not deny that the two are inherently linked. When my hypervisor is not stable, the applications suffer.
Reboots and Msft always go hand-in-hand. Msft shd basically rewrite a kernel specific for virtualization workloads just like VMware did. Else, they will keep crying over spilt milk.