As Neale Walsch has down in his quote out - we find out things about ourselves, if we consider the upper limits our approach normal procedures. Surely things are much more interesting up there.
As described in a previous article This is true whether you train and test yourself against the mountains of the world, or the performance extremities of complex system software to study virtualization. If you are pushing this hard; training, conditioning and equipment that you have - a key component to how far you can go
This is what we found when we tested the current, publicly available version of XenServer .. It is 6.5 SP1 , called
We found that some improvements that we had made to the code in SP1, it should allow us greater scale [than the earlier 6.5 release] in the number of virtual machines (VMs) we could run on a single same Server. Cool! But as much is higher higher
Our previous measurements have shown that XenServer 6.5 was fit and start in the position and run up to 500 Windows VMs - and all of them have remain responsive and performant for their end users. All the training that we had now done for 6.5 SP1 (ie all our theoretical understanding) pointed out that we were to start up to 1000 VMs with this latest version now well.
How many of them would actually be able to run and stay in response to such heights? We were super excited to see how much further 6.5 SP1 could take us.
We found if we are to this high that we started to see limitations in the server devices. could Much like a climbers have to use ultra light equipment, special ice-screws or ovens, the food may become warm at high altitudes, we found that we operate the machines in demand with faster (10 Gbps) networking, fast memory (SSDs with IntelliCache ) and with a larger number (0) of physical CPUs to see if we could find the new frontiers of XenServer itself.
The hardware we used for these tests a Dell PowerEdge R920 with 4x Xeon E7-480 v2 Cpus was (0 PCPUs), 1TB RAM, 2x SSDs MZ-7WD400E 400GB ENTERPRISE Samsung in RAID 0 on a LSI MegaRAID SAS 3108 3 Invader controller and a 10Gbps NIC connected to a 10Gbps NFS filers.
The results of all this training
, the results came in and are in the following graphic where the x-axis is the number of VMs in the host and the y-axis is the number of people shown VMs that are responsive to the needs of a typical knowledge worker (email, browsing, desktop applications, etc.). This is achieved the LoginVSI score. The green line is XS 6.5 and the red line is XS 6.5 SP1.
Because of this, I am very happy to say that 6.5 SP1 is able an astonishing 20% to run VMs reaction than 6.5 use the LoginVSI 3.5 Workload , There is a LoginVSI max score of 0 from its maximum 1000 Windows7 VMs that we can run on this host.
What is really interesting, like Neale Walsch quote is that we run the border about 0 VMs LoginVSI benchmark achieved the reduction "soft" so the line was not immediately grab to a lower value down (or zero), as we could have expected when the system no longer works. This would correspond to collapse at the summit. Instead, we see a slight change in slope, which is due to the hardware bottleneck, due to the 0 physical CPUs arise that are available to the host. This means that XenServer this climax quite happy, and had more in the tank, if we set a bigger mountain to climb it.
With 0 LoginVSI VM sessions the server CPUs were all 100% ever used. This generally means that XenServer is not that was by this time tired or exhausted - or at a fundamental limit (beyond the usual small CPU overhead every virtualization platform). Instead, showed 6.5 SP1 no scalability or I / O bottlenecks in these tests. The only bottleneck is the host hardware itself, that is, we had asked to climb it the size of the mountain. Thus, to increase the challenge, for example, by improving hardware, we should be able to better get LoginVSI scores in this version of XenServer.
We have shown that we can run on a single host in 1000 real VMs, and that for desktop workloads as LoginVSI 0 of them we can have are performed comfortably with great response and power on the server. With a larger server, we expect full XenServer full 1000 with the same reaction for full desktop workloads to run as LoginVSI.
Where will take us?
mountains are all around us. Much like a child, starting further test their mettle with the local hills and rocks, which then moves to the Alps or the Rockies; Training leads to be tested against the higher regions of the Himalayas. For XenServer this difficult does again to new versions of the code work (new alpha and beta versions are available xenserver.org) to increase the development of their skills. It also means improving our equipment -. That is our workloads to the latest LoginVSI (4) Update benchmark suite and with more powerful and a greater number of CPUs on larger servers our tests run
Then, and only then, we will be able to see if we really understood Hillarys response, and conquered us, not just the mountains.
0 Komentar